Anthropic Threat Report Details Blocked Biological Weapons Research Attempts

The findings, released on Thursday, detail cases of concern detected over the past eight months. In addition to the biological weapons-related research, Anthropic said its AI model Claude was used by actors linked to a Russia-based cyber espionage campaign and by an Iranian propaganda institution. Other flagged activity included fake dating apps, hotel Wi-Fi scams, and surveillance tools built to identify dissidents.

The report comes amid growing debate over the risks posed by advanced AI. US Senator Bernie Sanders has called for a pause on advanced AI development and a ban on artificial superintelligence. President Donald Trump has downplayed such fears, saying on Thursday he was concerned that failing to win the AI race would put the US in a bad position.

Separately, Google reported on Tuesday that a person had attempted to use its Gemini AI tool to obtain a complete, step-by-step technical guide for synthesising weaponised biological agents. Anthropic also accused Chinese AI firms of engaging in malicious activity, though the report did not provide details.

The company's threat intelligence report is part of a growing trend among AI firms to publicly disclose misuse of their platforms.

Why This Matters

The report provides concrete evidence that advanced AI models are being targeted for dual-use research with potential catastrophic consequences.
It highlights the challenge for AI companies: distinguishing between legitimate scientific research and malicious intent.
The disclosure fuels an ongoing policy debate about whether to pause or accelerate AI development.

Background

Anthropic is a leading AI safety company, known for its Claude model. The company has long warned about the potential for AI to be misused for bioterrorism or other harmful purposes. The threat intelligence report is part of its commitment to transparency and proactive monitoring. Similar reports have been published by other AI firms, including Google and OpenAI, as the industry grapples with growing misuse.

Key Perspectives

Anthropic: The company argues that proactive monitoring and disruption are necessary to prevent harm, but acknowledges the difficulty of determining intent. Policymakers: Some, like Senator Bernie Sanders, view the incidents as evidence that the technology is advancing too quickly and requires a moratorium. Others, like President Trump, prioritize competitive advantage over precaution. Critics/Skeptics: Some researchers argue that such reports may overstate risks or that the blocked research could have been legitimate scientific work. There is also concern that security measures may stifle beneficial research.

What to Watch

Whether other AI firms report similar biological weapons-related attempts in their upcoming threat intelligence reports.
Potential legislative action in the US Congress following the report's release.
How Anthropic and others refine their screening processes to avoid blocking legitimate research.

ZOTPAPER