AI USED FOR POTENTIAL BIOWEAPONS DEVELOPMENT? ANTHROPIC SAYS IT BLOCKED 5 ATTEMPTS

Fri 11 September 2026:

Company says users sought assistance involving chikungunya, avian influenza and smallpox

Artificial intelligence company Anthropic said Thursday it had blocked five attempts to use its AI models for activities that could potentially aid the development of biological weapons.

“While evaluations are useful because they provide evidence of capability, they cannot concretely demonstrate that such capability would ever be used to develop biological weapons in the real world,” the company said in a report examining the misuse of its AI systems.

According to the report, the cases involved research connected to chikungunya, avian influenza and smallpox.

__________________________________________________________________________

https://whatsapp.com/channel/0029VaAtNxX8fewmiFmN7N22

__________________________________________________________________________

Researcher Sought Help With Chikungunya Study

In one incident, a researcher linked to an unidentified military institution used Anthropic’s Claude chatbot while developing a research proposal involving chikungunya, a virus transmitted by mosquitoes.

While such research can have legitimate objectives, including developing vaccines, the researcher also requested assistance that raised concerns about potentially making the virus more dangerous.

Other cases involved attempts to use AI to investigate methods for increasing the transmissibility of avian influenza, as well as collecting information about animal toxins that could potentially have weapons applications.

Anthropic also reported that some users attempted to circumvent its safety measures through fraudulent accounts, private networks and third-party resellers.

AI Misuse Extended Beyond Biological Research

The company said its investigation uncovered other potentially harmful uses of AI, including efforts related to conventional weapons and cyberattacks.

One case involved queries originating from Yemen about developing a guided rocket using mobile technology, according to the report.

However, Anthropic stressed that detecting potentially dangerous capabilities does not by itself prove that users intended to develop biological weapons or that they would have been capable of successfully doing so.

The findings come as concerns grow over whether increasingly sophisticated AI systems could make it easier to carry out harmful activities that previously required highly specialized knowledge.

Anthropic said it has strengthened safeguards covering sensitive fields, including biology and cybersecurity. Access to some of its most advanced biology-focused models is also limited to organizations that have undergone vetting.

SOURCE: INDEPENDENT PRESS AND NEWS AGENCIES

__________________________________________________________________________

FOLLOW INDEPENDENT PRESS:

WhatsApp CHANNEL
https://whatsapp.com/channel/0029VaAtNxX8fewmiFmN7N22