AI company Anthropic says it has blocked attempts to use its models to conduct cyber attacks and research potential biological weapons, prompting experts to call for stricter access controls.
Key facts
- Anthropic says it blocked efforts to use its AI models to research potential biological weapons.
- The company also says it blocked attempts to use its models to carry out cyber attacks.
- The disclosures appear in Anthropic's threat intelligence report.
- Experts are urging stricter access to AI models amid rising reports of misuse.
- The report follows a warning from a former top Anthropic researcher about AI risks to humanity.
Anthropic, the company behind the Claude AI models, says it has blocked efforts to use its artificial intelligence to carry out cyber attacks and to conduct research that could have led to the creation of biological weapons, according to reporting by Sky News and other outlets.
The disclosures were set out in the company’s threat intelligence report, the BBC reported. The report details cases in which Anthropic says it identified and stopped attempts to misuse its systems, including for biological research risks and offensive cyber activity.
According to Al Jazeera, the report points to a rising number of misuse cases, with biological research among the risks flagged. The outlet said experts are urging stricter access to AI models in response to the findings.
Sky News reported that the blocked efforts spanned both weapons development research and cyber attacks, indicating the company is monitoring for a range of potential abuses of its technology.
The BBC noted that the revelations come after a former top researcher at the company warned of the risks that AI could pose to humanity, adding to a wider debate about how powerful AI systems should be controlled.
The sources provided do not include specific details on who attempted the misuse, when the attempts took place, or the exact methods Anthropic used to detect and block them.
Why it matters
The case highlights growing concern that advanced AI tools could be misused to aid dangerous activities, from cyber attacks to weapons research. It also puts a spotlight on how AI companies police their own systems and whether tighter access controls are needed to prevent harm.
Frequently asked questions
What did Anthropic block?
According to Sky News, Anthropic says it blocked efforts to use its AI models to carry out cyber attacks and to conduct research that could have led to the creation of biological weapons.
Where were these findings reported?
The disclosures appear in Anthropic's threat intelligence report, as reported by the BBC, Sky News and Al Jazeera.
What are experts calling for?
Al Jazeera reported that experts are urging stricter access to AI models amid rising cases of misuse, including biological research risks.