Penicillin: Discovery, History, Antibiotic Resistance & Future

 The Mold That Saved Millions — and the Race to Save Antibiotics

A chance discovery in 1928 changed medicine. Now resistant bacteria are forcing scientists to rethink the next century of treatment.

In 1940, a doctor could take an X-ray, give a blood transfusion and perform surprisingly sophisticated surgery. Some infections could already be treated with sulfa drugs. But if a dangerous bacterium entered the bloodstream or took hold deep in a wound, the choices were still frighteningly limited. Pneumonia could overwhelm a healthy adult. A mother might survive childbirth only to die of an infection afterward.

Then came a treatment so effective that people called it a miracle. Its story begins with a patch of mold in a London laboratory. But the most surprising part is not the accident that revealed penicillin. It is what happened next: a decade of stalled progress, an early patient who improved before the medicine ran out, a wartime manufacturing race, and a revolution that brought a new evolutionary challenge.

Nearly a century later, the question has changed. We no longer ask whether microbes can produce lifesaving chemicals. We ask how long our best treatments will remain effective—and whether new chemistry, artificial intelligence and even bacteria-killing viruses can help us stay ahead.

In 1928, Alexander Fleming noticed that an ordinary mold could stop bacterial growth. His discovery helped launch the antibiotic era and changed the future of medicine.

Before the miracle: when infection was the real enemy

People sometimes describe the era before antibiotics as a time when every scratch was a death sentence. That is too dramatic. Most minor injuries healed, and the immune system has always been capable of defeating many infections. The terrifying difference was uncertainty. A wound that looked manageable on Monday might become uncontrollable by Thursday, and doctors had few reliable ways to change the outcome.

By the early twentieth century, germ theory and antiseptic surgery had transformed medical practice. Surgeons understood why sterilizing instruments and cleaning wounds mattered. Vaccines could prevent some infections. Researchers had also developed certain targeted antimicrobial treatments. In the 1930s, sulfonamide drugs provided important treatment options. Penicillin was not the first substance ever used against bacteria.

Yet serious bacterial disease remained a formidable danger. Streptococci could invade wounds and blood. Staphylococci caused abscesses and infections following surgery. Certain forms of pneumonia, meningitis and puerperal fever remained deadly. Even a successful operation could be undone by infection. There was a gap between preventing contamination and treating infection after it took hold.

The eventual answer would come from something that had been growing unnoticed around us all along.

1928: Alexander Fleming notices something strange

Alexander Fleming, a Scottish bacteriologist at St Mary's Hospital in London, studied staphylococci. His laboratory contained stacks of bacterial cultures, and in 1928 one of those cultures became contaminated with a mold. Around the colony, bacterial growth had been inhibited, leaving a clear zone on the plate.

Fleming did not watch bacteria melt before his eyes or instantly invent a drug. Instead, he saw something worth investigating: the mold appeared to release a substance that stopped nearby bacteria from growing. He tested the effect against different microorganisms and named the active substance penicillin after the fungal genus Penicillium. The original fungus was long described using older names such as P. notatum; a later taxonomic study identified Fleming’s strain as Penicillium rubens.

It is tempting to tell this as pure luck. But laboratories encounter contamination all the time. The scientific achievement was recognizing that an apparently spoiled experiment contained useful evidence. Fleming had already studied natural antibacterial substances, including lysozyme. He was prepared to ask the right question.

His findings appeared in 1929. The work was promising, yet it did not immediately change medical practice. Penicillin was unstable in crude preparations, difficult to isolate, and extraordinarily hard to manufacture in useful quantities. Fleming's research established an effect, not a readily available medicine. As the American Chemical Society's historical account makes clear, the journey from discovery to treatment depended on several people and institutions.

One popular version of the story suggests Fleming returned from a holiday and instantly understood that he had discovered a medical revolution. The actual history was more gradual. He had found a powerful lead without the techniques, resources or production system needed to exploit it.

A century later, a computer can help identify a promising antibacterial molecule in days. Turning that molecule into something doctors can safely prescribe may still take years. Fleming’s story is a useful warning against confusing the first step with the finish line.

The Oxford team turns an observation into medicine

In the late 1930s, researchers at the University of Oxford returned to penicillin. Howard Florey, Ernst Chain, Norman Heatley, Edward Abraham and their colleagues approached the problem from several angles: microbiology, chemistry, purification, dosing and production.

Their laboratory was part research institute, part improvised factory. The mold needed space to grow. The useful compound had to be separated from a messy culture liquid. Each experimental batch was precious, and waste could mean the loss of a potential treatment.

On 25 May 1940, Oxford researchers tested penicillin in mice infected with dangerous bacteria. The treated mice survived, while the untreated controls died. It was strong evidence that penicillin might fight an infection inside the body—not merely inhibit bacteria in a laboratory dish. The Oxford historical timeline records this turning point.

The next test was far harder. A living patient needed much more medicine than a small group of mice.

On 12 February 1941, doctors began treating Albert Alexander, a 43-year-old policeman gravely ill with a staphylococcal infection. His condition improved noticeably during treatment. The team had evidence that the drug could work inside the human body. But they did not have enough of it. In a desperate effort to stretch supplies, researchers recovered penicillin from his urine for reuse. Eventually the drug ran out, the infection returned and Alexander died.

Alexander is often called the first patient treated with penicillin. Strictly speaking, earlier local applications had been attempted; his case was a landmark in the Oxford team’s efforts to use purified penicillin for a life-threatening infection throughout the body.

Alexander’s improvement had shown the doctors what might be possible. His death showed them what was still missing: enough penicillin to complete a course of treatment.

In the early 1940s, Howard Florey, Ernst Chain, Norman Heatley, and their colleagues transformed Fleming's discovery into a practical treatment, laying the groundwork for mass-produced antibiotics.

Wartime urgency, American factories and a moldy melon

World War II brought urgency, money, industrial chemistry and international collaboration to the problem. Oxford researchers sought help in the United States. At the US Department of Agriculture's laboratory in Peoria, Illinois, scientists investigated how to grow mold more efficiently and increase production.

The breakthroughs were remarkably practical. A by-product of corn processing, known as corn steep liquor, proved to be excellent food for the mold. Improved nutrients, aeration and deep-tank fermentation helped replace small surface cultures with industrial production. Chemists and engineers made something previously obtainable by the milligram available on an enormous scale.

They also searched for better-performing strains. One exceptionally productive Penicillium strain came from a moldy cantaloupe associated with a Peoria market. The fruit became part of scientific folklore. You may encounter a confident story that a technician named Mary Hunt found the melon. She contributed to the search, but the USDA's own historical investigation notes that the exact identity of the person who found it remains uncertain.

The unresolved identity of the person who found the fruit is a reminder of how many hands contributed to the breakthrough. Scientists selected better strains, engineers built fermentation systems, and factory workers helped turn a scarce laboratory material into reliable doses.

By the Allied landings in Normandy in June 1944, wartime production was large enough for penicillin to be used in treating wounded personnel. Civilian access expanded after the war. In 1945, Fleming, Florey and Chain shared the Nobel Prize in Physiology or Medicine. The public remembered three names; the scientific achievement belonged to a much wider group of researchers, technicians and production workers.

The change reached far beyond treating infected wounds. Antibiotics helped make more ambitious surgery possible and later became essential support for treatments that leave patients vulnerable to infection, including chemotherapy and organ transplantation. Better hygiene, anesthesia, intensive care and other innovations mattered too; antibiotics were one powerful part of a much larger medical transformation.

What does penicillin actually do inside a bacterium?

To understand both its success and its limitations, imagine a bacterium as a tiny pressurized cell surrounded by a protective mesh. Many bacteria have a wall made of peptidoglycan, a network that helps them keep their shape instead of bursting under internal pressure.

Penicillin belongs to the beta-lactam family of antibiotics. It blocks enzymes known as penicillin-binding proteins, which help link the wall’s peptidoglycan strands together. When a susceptible bacterium grows and attempts to build new wall material, penicillin disrupts the process. The damaged structure can no longer do its job properly, and the bacterium may die.

Human cells do not have peptidoglycan walls. That difference creates a selective target: the drug interferes with something bacteria need but human cells do not possess. Selective, however, does not mean harmless. Penicillin can cause allergic reactions and other adverse effects, and not every bacterium is susceptible.

There are other limitations. Bacteria differ in the structure of their cell envelopes. Some have an additional outer membrane that makes entry difficult. Others produce enzymes, called beta-lactamases, that break down beta-lactam antibiotics. Still others change the proteins to which these drugs attach.

Most importantly, penicillin does not treat viruses. A flu virus or the virus behind a common cold does not construct the bacterial cell wall that penicillin attacks. Confusing bacterial and viral infections is one reason antibiotics are still used unnecessarily. For more on how viruses change rather than “plan” their evolution, see Next Horizon's explanation of viral mutations.

Penicillin blocks enzymes needed to build and maintain the bacterial cell wall. Without a properly formed protective structure, susceptible bacteria can weaken and die.

The golden age—and the beginning of resistance

After penicillin, scientists discovered or developed many other antibiotic families, including streptomycin, tetracyclines, macrolides and cephalosporins. Synthetic chemistry also expanded what medicine could do: researchers could modify natural molecules, improve their stability, change which bacteria they affected and work around some resistance mechanisms.

But microbes were not passive opponents. Bacteria reproduce quickly, populations are enormous, and genetic variation is constant. Under antibiotic pressure, susceptible organisms may be eliminated while resistant ones survive. Resistance can also spread when bacteria exchange genetic material. A resistant strain can emerge through mutation or acquire resistance genes from other bacteria.

This is evolution, not intention. A bacterium does not decide to “learn” an antibiotic or become more intelligent. Variants and genes that happen to survive a particular environment become more common. A drug that works wonderfully in one decade may face a very different microbial population in the next.

Fleming recognized the risk early. In his 1945 Nobel lecture, he warned that exposing bacteria to too little penicillin could select resistant microbes. That warning has grown into a much broader concern: resistance is shaped by antimicrobial use across healthcare, agriculture and the environment, while infections also spread through failures of sanitation and infection control.

Resistant infections do not mean that every antibiotic is useless. They mean treatment becomes less predictable, more expensive, slower and, in some cases, impossible with the medicines available locally. A severe infection can become a race between laboratory identification, clinical judgment and bacterial growth.

How serious is the antibiotic resistance problem in 2026?

The numbers are serious. They are also easy to misread. In its 2025 global surveillance report, the World Health Organization estimated that one in six laboratory-confirmed bacterial infections included in its analysis in 2023 showed resistance to antibiotic treatment. This was not a claim that one in six infections of every type, in every location, had become untreatable. The figure covers selected common infections and antibiotics, using surveillance data with important regional gaps.

A 2024 study in *The Lancet*, produced by the Global Research on Antimicrobial Resistance collaborators, estimated that bacterial antimicrobial resistance could be directly responsible for about 1.91 million deaths in 2050 under its reference scenario. Across 2025–2050, its projected cumulative total was about 39 million deaths attributable to resistance. Those are modelled scenarios, not deaths that have already occurred, and the estimates can change as treatment, prevention and surveillance change.

The distinction between attributable to and associated with resistance also matters. The first asks which deaths may not have occurred if the infection had been drug-susceptible; the second counts deaths involving drug-resistant infections, a broader category. Mixing the two creates eye-catching but misleading headlines.

The threat is not confined to one “superbug.” Hospitals worry about resistant Acinetobacter baumannii, Klebsiella pneumoniae, certain E. coli, resistant Staphylococcus aureus and other organisms. Their clinical behavior differs. A drug that helps against one may do little against another. Local laboratory data can be as important as the drug's reputation.

And there is a second problem, almost the opposite of overuse. In poorer healthcare systems, patients may not have timely access to diagnostics or even to effective existing antibiotics. The challenge is to stop unnecessary treatment and make necessary treatment available. Both are essential.

The discovery pipeline is thinner than it looks

New antibiotics still reach patients. But an impressive-looking list of experimental compounds can create a false sense of security. Every candidate must survive many filters: Is it toxic? Does it reach infected tissue? Does the body clear it too quickly? Can it be manufactured consistently? Does it actually outperform available treatment in people? Will resistance appear almost immediately?

The WHO's October 2025 pipeline assessment listed 90 antibacterial candidates in clinical development, down from 97 in the earlier comparison. Of those, 50 were traditional antibacterial drugs and 40 used non-traditional approaches, such as phages or antibodies. Only 15 met the report's innovation criteria; only five showed activity against at least one pathogen in the WHO's critical-priority group. These categories describe the pipeline under review, not an inventory of all approved antibiotics.

The same report recorded 232 preclinical programmes. That sounds encouraging, and in some ways it is. But a laboratory concept is several steps away from a licensed medicine. Some molecules fail because they do not work in animals, some because they harm human cells, and others because production or clinical trials prove too difficult.

In March 2026, WHO went further and published target profiles for antibiotics the world most needs: drugs for serious resistant Gram-negative infections, resistant Gram-positive infections in highly vulnerable patients, and bacterial meningitis. The goal was not merely “more antibiotics.” It was treatments that are effective, safe, practical, affordable and accessible.

That shifts the future story from a race to collect molecules to a challenge of designing an entire medical system.

Antibiotic resistance is one of modern medicine's greatest challenges. Scientists study bacterial growth and resistance patterns to understand which treatments remain effective.

New targets: antibiotics that attack bacteria differently

New molecular targets. Some antibiotics are chemical variations on earlier families. Those modifications can be valuable, but resistance mechanisms sometimes carry over. Researchers therefore look for vulnerable pieces of bacterial machinery that previous drugs have barely touched.

A striking example is lariocidin, a naturally produced, knot-like peptide reported in Nature in 2025. The molecule interferes with the bacterial ribosome—the machinery that builds proteins—but binds at an unusual site. Researchers found activity against several pathogens and encouraging results in a mouse infection model. The original study was exciting precisely because it suggested a fresh chemical starting point. It did not show that lariocidin was already safe and effective in human patients.

Another experimental approach, zosurabalpin, targets the transport of molecules needed to construct the outer envelope of A. baumannii. Its significance lies in attacking a hard-to-treat Gram-negative organism through a different mechanism. Promising animal results and early clinical investigation do not guarantee final approval, but they widen the range of targets available to researchers.

Real approvals, not only laboratory hopes. It is equally important to notice genuine progress. In March 2025, the US Food and Drug Administration approved gepotidacin for specified uncomplicated urinary tract infections in eligible female patients. In December 2025, the FDA approved zoliflodacin for uncomplicated urogenital gonorrhea in eligible patients. These drugs do not solve antibiotic resistance across the board; each has a defined indication and eligibility rules. They demonstrate, however, that the innovation pipeline has not stopped entirely. Their status is documented in the FDA's gepotidacin and zoliflodacin trial summaries.

One lesson repeats across these examples: the goal is increasingly to match a specific molecule to a specific bacterial vulnerability, rather than expecting a single universal drug to cover every infection.

AI can suggest new antibiotics. Can it help deliver them?

For much of antibiotic history, discovery involved collecting compounds and testing them against bacteria. Scientists still do this, and nature remains a remarkable source of molecules. But there are far too many possible chemical structures to test them all in a laboratory.

Machine learning changes the search process. A model can examine molecular patterns associated with antibacterial activity and rank compounds that chemists might otherwise overlook. More recently, generative systems have begun proposing molecules that are not simply retrieved from existing chemical catalogs.

In 2020, researchers reported using deep learning to identify halicin, a molecule with antibacterial activity and encouraging results in mouse models. In 2023, another team described abaucin, a narrow-spectrum candidate active against Acinetobacter baumannii in preclinical work. These were not instant AI-created medicines for pharmacy shelves. They were research leads obtained by combining computational prediction with conventional experiments. See the original studies in *Cell* and *Nature Chemical Biology*.

Then a 2025 Cell paper raised the ambition. Researchers used generative methods to explore enormous numbers of fragments and proposed chemical structures against drug-resistant Neisseria gonorrhoeae and Staphylococcus aureus. They synthesized 24 candidate compounds, of which seven showed selective antibacterial activity. Two leads showed efficacy in mouse models, using mechanisms different from those of currently used antibiotics. These are impressive experimental results, not completed human trials. The peer-reviewed study makes both the possibilities and the stage of development clear.

AI's value is not that it replaces microbiologists. It can make the search less blind. Chemists must still synthesize molecules; laboratories must test them; clinicians must verify safety and clinical benefit. The model can be wrong about toxicity, absorption, interactions and resistance. It can also favor chemicals that look statistically promising but are impractical to manufacture.

A useful analogy is a map of an enormous, unexplored library. AI may point to shelves nobody had time to inspect. Human science still has to open the books and check whether the instructions work.

This is closely related to the broader story of AI and genetic research: algorithms are becoming powerful guides to experiments, not substitutes for biological evidence.

Bacteriophages: the viruses that hunt bacteria

Long before the antibiotic era, doctors experimented with another way of fighting infection: viruses that infect bacteria.

These viruses are called bacteriophages, or simply phages. A phage attaches to a suitable bacterial cell and introduces genetic material; some phages then multiply inside the bacterium and break it open. This lytic behavior is central to most proposed phage treatments. Because many phages are highly selective, researchers can potentially attack a pathogen while leaving much of the surrounding microbiome untouched.

This specificity is both their promise and their difficulty. The phage that kills one bacterial strain may fail against a closely related strain. The human immune system can affect treatment. Bacteria can develop resistance to phages, too. Producing standardized, stable and safe phage preparations is not trivial. Delivering them to a lung, a bloodstream infection or a protective biofilm raises different questions.

Phage therapy has never entirely disappeared. It continued in countries including Georgia and Poland even as Western medicine largely shifted to antibiotics. But historical use and individual rescue stories are not substitutes for strong results from well-designed clinical trials. The history of phage medicine is fascinating; its modern evidence base is still developing. A 2026 review in *Nature Reviews Microbiology* emphasizes how difficult it remains to predict clinical success from laboratory susceptibility tests.

Still, the field is moving. In September 2026, an international group published a consensus guideline in *Nature Medicine* with more than 60 recommendations for personalized phage therapy. That is an important step toward rigorous selection, quality control, patient monitoring and research standards. It does not mean phages have become a routine, universally proven alternative to antibiotics.

In a specialist hospital of the future, a patient’s bacterial strain might be matched against a carefully maintained collection of therapeutic phages when standard drugs fail. The fascinating possibility is adaptive medicine: treatment designed around the actual microbe in the patient, rather than an average microbe described in a textbook.

But matching, timing and manufacturing would have to become much faster and more reliable for that future to reach ordinary healthcare.

CRISPR, precision drugs and the microbes we want to protect

What if a treatment could recognize a particular resistance gene or bacterial strain and target it while sparing its neighbors? That is the promise behind CRISPR-based antimicrobials, often delivered through engineered phages or other carriers. In experimental systems, molecular tools can target selected bacterial DNA sequences. The difficulty is getting those tools into enough bacteria safely and predictably, in the right tissue, at the right time.

This is closely connected with bioengineering and synthetic biology. But programmable does not mean clinically ready. Delivery, off-target ecological effects, changes in microbial communities and the possibility of new resistance all require testing. CRISPR-based antibacterial treatments remain experimental; no widely approved CRISPR therapy is available for routine bacterial infection care.

Researchers are also reconsidering a basic assumption of antibiotic therapy: perhaps the best drug is not the one that kills the broadest range of bacteria. The human body hosts vast microbial communities, especially in the intestine, and many of those organisms help resist dangerous colonizers. Broad-spectrum antibiotics can disturb that ecosystem.

In a 2025 *Nature Communications* study, the experimental compound EVG7 reduced Clostridioides difficile infection and recurrence in a mouse model while sparing some gut bacteria associated with protection against reinfection. Again, mice are not patients and experimental success is not a regulatory approval. The concept is important: future antibiotics may be judged partly by how much useful microbiology they leave alive, not just by how many dangerous organisms they eliminate.

This could lead to more targeted antibiotic combinations, treatments that protect beneficial microbes, and therapies designed around the interaction between pathogen, microbiome and immune system. Precision medicine may mean more than choosing the right pill. It may mean treating an entire microbial ecosystem carefully.

The next generation of infection treatments may combine AI-designed antibiotics, rapid diagnostics, and bacteriophage therapy. Researchers are exploring these approaches to overcome drug-resistant bacteria.

The test that could matter as much as a new drug

Consider what happens when a severely ill person arrives at an emergency department. Doctors may suspect a dangerous bacterial infection, but they often cannot wait for a traditional culture to reveal the exact species and its vulnerabilities. They may begin broad treatment immediately because delay itself can be dangerous.

A faster diagnostic system could change that decision. Researchers are advancing rapid molecular tests, sequencing and improved methods for measuring how bacteria respond to drugs. The ideal is to identify a pathogen and useful treatment as soon as possible, rather than prescribing several broad options and narrowing only after laboratory results arrive.

There is no single universal test that can instantly identify every infection and every resistance mechanism. Detecting a resistance gene is not always identical to measuring whether a particular drug works in the patient. Samples can contain too few bacteria, mixed organisms or contaminants, and some infections are especially hard to access. Real-world cost and laboratory capacity matter as much as technical speed.

Getting the right antibiotic to a critically ill patient earlier can make an enormous difference. Faster testing could also reduce unnecessary broad-spectrum prescribing. That is why the WHO’s research agenda includes diagnostics, not only medicines. An intelligent diagnostic platform might eventually combine rapid lab signals with local resistance data to guide treatment—while leaving clinicians responsible for decisions.

The hardest problems are not all in the laboratory

Antibiotic resistance is often described as an arms race, as if the only available response were to invent a stronger weapon. But many of the biggest gains may come from preventing infections in the first place: vaccination, safer surgery, clean water, hospital infection control, sanitation, and rapid treatment of the patients who truly need it.

There is also a difficult economic contradiction. Developing an antibiotic is expensive and risky, but responsible healthcare tries to reserve especially valuable new drugs for cases in which they are needed. A company can spend years developing a successful product and then sell relatively few doses. That is often good clinical stewardship but a weak business model.

The United Kingdom has experimented with an NHS subscription-style payment model, linking payments more closely to the public health value of important antimicrobials rather than simply the quantity sold. Whether such payments can scale internationally is still debated. The underlying idea deserves attention: a life-saving antibiotic can have public value even when responsible doctors prescribe relatively little of it.

Any future strategy must also face an uncomfortable reality. A breakthrough that exists only in wealthy hospitals will not solve a global resistance crisis. A child needs an accessible existing antibiotic and a functioning clinic more urgently than a spectacular molecule still in a mouse study. Access, responsible prescribing, agriculture, environmental contamination and public health surveillance are parts of the same system.

A hospital in 2035: one possible future

Imagine a patient with a serious infection. Within hours rather than days, a laboratory identifies the organism and a useful resistance profile. Software helps clinicians interpret the results in the context of local outbreaks and the patient's history. For a routine susceptible infection, a familiar inexpensive antibiotic remains the best choice. For an unusually resistant infection, specialists consider a newer drug, a carefully selected combination or, in appropriate settings, a validated phage-based approach.

Behind the scenes, AI assists chemists in designing and screening new compounds. Hospitals share anonymized resistance trends. Public agencies invest in vaccines, water systems and infection prevention. Drug development companies are paid in ways that do not depend on encouraging unnecessary use.

Nothing about that picture requires a magic cure. It requires a set of improvements—some already in use, some experimental, and some limited by economics and infrastructure. No responsible scientist can guarantee that all will arrive by 2035. But it is a more credible future than a headline promising that AI will defeat every superbug next year.

There are also less hopeful possibilities. Without better stewardship, access and discovery, more infections could become complicated and costly to treat. Resistance may rise faster in some regions than in others. New treatments themselves will eventually create new evolutionary pressures. Progress could be real without ever becoming permanent victory.

Evolution will keep happening. The realistic aim is to detect resistance sooner, prevent avoidable infections, and use existing medicines wisely while developing better ones.

The lesson of the moldy dish

Fleming noticed a clear ring around a colony of mold. The Oxford team discovered how to turn that observation into a treatment. Industrial researchers and workers found a way to produce it at scale. It took more than one kind of intelligence to make the breakthrough matter.

That is why penicillin's story matters far beyond its history. The next valuable treatment might come from a soil microbe, an AI-guided chemistry experiment or a carefully selected phage. The next life saved might come from a faster laboratory test. It may also come from a less glamorous decision: paying for infection-control staff, making an existing drug available, or prescribing an antibiotic only when it is truly needed.

Penicillin changed medicine because humans combined curiosity with chemistry, biology, engineering and an urgent reason to act. Preserving its legacy will demand the same combination—plus a new respect for the bacteria that never stopped evolving.

The first breakthrough began with an accident. Making sure antibiotics still work for the next generation will take deliberate choices.

Frequently asked questions

Who discovered penicillin? Alexander Fleming identified its antibacterial effect in 1928. Howard Florey, Ernst Chain, Norman Heatley, Edward Abraham and many colleagues later helped turn the finding into an effective treatment; scientists and industry developed large-scale production.

When was penicillin first used to treat people? Crude preparations had some earlier local uses. The Oxford team began its famous early systemic treatment of Albert Alexander in February 1941.

Does penicillin still work today? Yes, against infections caused by susceptible bacteria. Some bacteria are resistant, and penicillin is not the right choice for every bacterial infection.

Can antibiotics treat the flu or a common cold? No. Flu and the common cold are caused by viruses. Antibiotics do not work against those viruses, though a separate bacterial complication may sometimes require treatment.

Will AI replace antibiotics? AI can help find and design new antibiotic candidates, but it is a research tool, not a treatment. Proposed molecules still need laboratory experiments, clinical testing and regulatory review.

Could bacteriophages replace antibiotics? In some settings they are already used, usually through specialized or individualized pathways. They may become useful alternatives or partners for selected infections, but as of October 2026, broader routine use still faces evidence, manufacturing and regulatory hurdles.

Are we heading back to the pre-antibiotic era? Not inevitably. Resistance is a serious problem, but prevention, reliable diagnostics, responsible antibiotic use, new drugs and carefully validated alternatives can reduce its impact. Future outcomes depend on choices made now.