Health and Me

Rachel V Thomas

Anthropic Says Claude Blocked Bird Flu, Chikungunya Research: How AI Could Enable Biological Misuse

Anthropic Says Claude Blocked Bird Flu, Chikungunya Queries: How AI Could Enable Biological Misuse

In an alarming report, Anthropic, an artificial intelligence safety and research company and developer of the Claude family of large language models, said it blocked multiple accounts that used its AI models in ways that could support the development of biological weapons.

In a report published Thursday, the San Francisco-based AI company described biological misuse as one of the “most serious risks” to increasingly capable AI models. The report covers attempts to misuse Claude between December 2025 and August 2026, including activity involving biology, weapons, cyberattacks, influence operations and surveillance.

It details five cases involving potentially concerning biological research, including <a href="https://www.healthandme.com/health/chikungunya">chikungunya</a> gain-of-function research, avian influenza, orthopoxviruses, venoms and toxins.

“We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them,” Anthropic said on X.

Also read: <a href="https://www.healthandme.com/health-news/ai-creates-16-viruses-to-kill-drug-resistant-e-coli-raises-biosafety-concerns-article-155382934" data-type="tilCustomLink">AI Creates 16 Viruses To Kill Drug-Resistant E. coli, Raises Biosafety Concerns</a>

Anthropic identified about 35 potentially concerning “research efforts” over 30 days. It could not determine whether users “intended harm” or were conducting legitimate research.

Five cases involved users circumventing safeguards or regional restrictions:

<ul><li>Chikungunya: AI was used in a gain-of-function project after safeguards were bypassed.</li><li>Avian influenza: AI-assisted planning was used for experiments aimed at adapting the virus to mammals.</li><li>Orthopoxvirus: AI was used to generate a grant application on viral immune evasion.</li><li>Venoms: AI was used to optimise venom-related molecules.</li><li>Toxins: AI was used to computationally redesign toxins.</li></ul>Anthropic said increasingly capable AI could raise misuse risks and that newer models have stronger safeguards for dual-use biological research.

<h2>Anthropic Says It Stopped The Operations</h2>

“We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards,” Anthropic said.

The company said it also shared information with authorities and other AI companies where appropriate.

“These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen,” it said.

<h2>Can AI Enable Biological Misuse?</h2>

Read More: <a href="https://www.healthandme.com/health-news/ai-designed-viruses-have-been-successfully-tested-in-lab-but-scientists-are-worried-heres-why-article-155409395" data-type="tilCustomLink">AI-Designed Viruses Have Been Successfully Tested In Lab But Scientists Are Worried; Here's Why</a>

The report comes as former Anthropic employee Jacob Coxon resigned this week, stating on X that “the people building AI earnestly believe that it could kill us all by the end of the decade.”

The 27-year-old AI researcher said it could hack critical infrastructure — or build “extinction-level bioweapons.”

“There’s a lot of ways that the AI could actuate itself in the world,” he said. “If you extrapolate into the future, the level of capabilities of these AIs … they could cause extreme havoc,” Coxon later told CNN.

In July, nearly 1,400 AI company employees signed an open letter urging the US government to regulate the technology and slow its development to ensure its safety.

OpenAI Chief Scientist Jakub Pachocki also warned this week that AI capabilities are advancing faster than researchers’ ability to reliably monitor and control them.

Dr Ashish K. Jha, physician, scientist and biosecurity expert, said the Anthropic cases show how AI could lower some barriers to biological research.

“We are sleepwalking into a potentially huge disaster. And it amazes me how little attention it is getting,” Jha said on X.

“Anthropic just published five case studies today of people using Claude in ways that could have helped develop biological weapons. They banned the accounts, traced the networks, and told other AI labs and governments what they found.”

Jha said the barriers that once made biological weapons difficult to develop included “deep expertise, advanced equipment, years of specialized training.” “Those barriers are dissolving. Rapidly,” he said.

Jha also pointed to a case in which Claude blocked a request, after which the operator built a system to route it to a more permissive model.

“In a world of lots of open access models that are getting quite good, this will keep happening,” he said.

Health Conditions A-Z

health

anthropic-says-claude-blocked-bird-flu-chikungunya-queries-how-ai-could-enable-biological-misuse

Anthropic

Claude

Bird Flu

AI bioweapon

Jacob Coxon

Chikungunya

taking-atorvastatin-what-happens-if-you-stop-the-cholesterol-drug

Taking Atorvastatin? What Happens If You Stop The Cholesterol Drug

us-shelved-poultry-vaccination-due-to-rfk-jrs-vaccine-concerns-can-vaccine-help-control-bird-flu

US Shelved Poultry Vaccination Due To RFK Jr’s Vaccine Skepticism: Can Vaccine Help Control Bird Flu?

pope-leo-xivs-lung-cancer-diagnosis-when-should-you-worry-about-lung-nodules

Pope Leo XIV’s Lung Cancer Diagnosis: When Should You Worry About Lung Nodules?

Health News

8 Sit-Up Variations to Strengthen Your Core and Build Stability

Benefits Of Ayurvedic Health Supplement Ashwagandha

9 Anti-Inflammatory Breakfast Options You Should Try

5 Foods You Should Avoid For PCOS, Says Gynecologist

Love Sitting With Your Legs Crossed? Here’s You Should Avoid It

6 Daily Foods That Are Secretly Increasing Your Calories

9 Brain Boosters You Should Do Everyday

Hip Exercises That Will Keep You Flexible As You Age

9 Leg Exercises Office Workers’ Should Do At Home

Benefits of Having Herbal Tea Every Morning

9 Daily Habits That Boost Your Brain Power

Lower-Body Exercises That Are Game-Changers

How Drinking Apple Cider Vinegar Helps You

Aging Myths You Should Never Believe

Superfoods That Support Your Gut Health

What Happens To Your Body When You Eat Instant Noodles Every Day?

Health Issues Premature Babies Often Face

Exercises Office Workers Should Do For Their Heart Health

Reasons Why You Should Add Raw Turmeric To Your Diet

Benefits Of Drinking Milk At Night: Why You Should Do It

Food That Provides Vitamin B12 Naturally In Your Diet

Surprising Health Benefits Of Fenugreek Seeds

9 Superfoods You Should Eat For Your Brain Health

What Does Mental Health Support Look Like At Work?

Long Stretches You Should Do To Unwind From Your Day

Colon Cancer In Young Adults: Early Symptoms You Should Know

Health Benefits Of Eating Dry Fruits Everyday

Are You At Risk For Fried Rice Syndrome?

These Tests Reveal How Well You Are Aging

Lower-Back Exercises You Should Do After A Long-Day's Work

Children’s Diet – What Kids Need To Eat For Immunity Building

Common Diabetes Myths That Could Hurt Your Health

Love Spicy Food? Here’s How It Affects Your Health

Superfood List: Veggies And Fruits You Must Have For Immunity

Foods People With IBS Should Avoid

Exercises To Improve Digestion

Easy Work Lunches You Can Make In The Morning

Hunger Vs Craving: Difference Explained In 60 seconds

These Things Can Cause Your Heart To Age Faster

High Carb Foods That Are Actually Healthy

In an alarming report, Anthropic, an artificial intelligence safety and research company and developer of the Claude family of large language models, said it blocked multiple accounts that used its AI models in ways that could support the development of biological weapons.In a report published Thursday, the San Francisco-based AI company described biological misuse as one of the “most serious risks” to increasingly capable AI models. The report covers attempts to misuse Claude between December 2025 and August 2026, including activity involving biology, weapons, cyberattacks, influence operations and surveillance.It details five cases involving potentially concerning biological research, including chikungunya gain-of-function research, avian influenza, orthopoxviruses, venoms and toxins.“We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them,” Anthropic said on X.Also read: AI Creates 16 Viruses To Kill Drug-Resistant E. coli, Raises Biosafety ConcernsWhat Did Anthropic Find?Anthropic identified about 35 potentially concerning “research efforts” over 30 days. It could not determine whether users “intended harm” or were conducting legitimate research.Five cases involved users circumventing safeguards or regional restrictions:Chikungunya: AI was used in a gain-of-function project after safeguards were bypassed.Avian influenza: AI-assisted planning was used for experiments aimed at adapting the virus to mammals.Orthopoxvirus: AI was used to generate a grant application on viral immune evasion.Venoms: AI was used to optimise venom-related molecules.Toxins: AI was used to computationally redesign toxins.Anthropic said increasingly capable AI could raise misuse risks and that newer models have stronger safeguards for dual-use biological research.Anthropic Says It Stopped The Operations“We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards,” Anthropic said.The company said it also shared information with authorities and other AI companies where appropriate.“These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen,” it said.Can AI Enable Biological Misuse?Read More: AI-Designed Viruses Have Been Successfully Tested In Lab But Scientists Are Worried; Here's WhyThe report comes as former Anthropic employee Jacob Coxon resigned this week, stating on X that “the people building AI earnestly believe that it could kill us all by the end of the decade.”The 27-year-old AI researcher said it could hack critical infrastructure — or build “extinction-level bioweapons.”“There’s a lot of ways that the AI could actuate itself in the world,” he said. “If you extrapolate into the future, the level of capabilities of these AIs … they could cause extreme havoc,” Coxon later told CNN.In July, nearly 1,400 AI company employees signed an open letter urging the US government to regulate the technology and slow its development to ensure its safety.OpenAI Chief Scientist Jakub Pachocki also warned this week that AI capabilities are advancing faster than researchers’ ability to reliably monitor and control them.Dr Ashish K. Jha, physician, scientist and biosecurity expert, said the Anthropic cases show how AI could lower some barriers to biological research.“We are sleepwalking into a potentially huge disaster. And it amazes me how little attention it is getting,” Jha said on X.“Anthropic just published five case studies today of people using Claude in ways that could have helped develop biological weapons. They banned the accounts, traced the networks, and told other AI labs and governments what they found.”Jha said the barriers that once made biological weapons difficult to develop included “deep expertise, advanced equipment, years of specialized training.” “Those barriers are dissolving. Rapidly,” he said.Jha also pointed to a case in which Claude blocked a request, after which the operator built a system to route it to a more permissive model.“In a world of lots of open access models that are getting quite good, this will keep happening,” he said.