Anthropic Says Scientists Used Claude for Research with Potential Bioweapons Risks
Anthropic said it banned accounts after finding five cases of scientists using Claude for research that could support biological weapons development.
- In a report released Thursday, Anthropic detailed seven misuse areas for its AI models, including five instances involving biological research that could support weapons development. The AI-maker emphasized that while usage was detected, intent remains unconfirmed.
- Actors circumvented controls and obfuscated research goals to evade safeguards, Anthropic said. Because biological research can be dual-use, the AI-maker struggled to distinguish between legitimate scientific inquiry and malicious attempts.
- Beyond biological concerns, Anthropic documented six scenarios involving software for "firearms, missiles, armed drones, bombs, and other munitions." Other cases involved cyber operations, surveillance, and influence campaigns designed to hide identities.
- The report arrived two days after Jacob Coxon resigned from Anthropic, warning that AI development is moving too quickly. Coxon cautioned that increasingly capable systems could eventually become difficult for humans to control.
- Prompting calls from Senator Mark Kelly, Washington figures are reacting, with President Donald Trump expressing concerns about maintaining an AI lead over China. Evan Hubinger warned that superintelligence alignment remains unsolved.
12 Articles
12 Articles
Anthropic Says US Adversaries Aimed Claude at Weapons Research
Anthropic PBC says its artificial intelligence model Claude has been misused in attempts to develop a wide range of potential military applications, including kamikaze drone swarms, missile navigation systems and biological weapons.
Anthropic Blocked 5 Possible Attempts to Research Bioweapons with Claude AI
A new report from Anthropic also includes instances of targeted influence operations, weapon production, scams, and the stealing of AI technologies. Anthropic’s latest threat intelligence report details seven ways the AI-maker has found its models being misused, including five instances of research related to biological weapons development in unnamed countries.The 154-page report dives deep into dangerous activities Anthropic found its tools bei…
American artificial intelligence developer Anthropic blocked five attempts to use his models to create dangerous viruses. Incidents were reported from the threat assessment report, Axios reports. Between December and August, the perpetrators tried to use the Claude language model to assist in biological weapons research. In two cases, users requested information from neuronets to enhance pathogens in laboratory settings.
US technology company Anthropic has uncovered and thwarted several attempts to use its artificial intelligence models to develop biological weapons, it said in a 154-page report outlining the most serious incidents this year where its Claude models were misused or attempted to be misused.
Coverage Details
Bias Distribution
- 40% of the sources lean Left, 40% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
