Artificial intelligence company Anthropic has revealed a series of attempts by bad actors to exploit its AI systems for potentially dangerous biological research, underscoring growing concerns about the intersection of advanced AI and biosecurity.
The disclosure comes as part of Anthropic’s latest threat intelligence and misuse report, which documents how various actors sought to leverage the company’s Claude AI models for activities ranging from cyber operations and surveillance to biological research that could carry significant safety risks.
According to the report, Anthropic identified several cases involving scientific research requests that raised concerns about potential biological weapons development or other forms of harmful biological misuse. The company said some users attempted to circumvent safety controls and obscure their true intentions while interacting with its AI systems.
One case Anthropic highlighted involved requests related to gain-of-function research on the chikungunya virus, a type of research that could potentially increase a pathogen’s transmissibility or ability to evade immune responses. The company classified the activity as a serious dual-use concern because scientific techniques intended for legitimate research can also be adapted for harmful purposes.
Anthropic said it blocked or disrupted the identified misuse attempts and strengthened safeguards to prevent its most capable models from assisting with dangerous biological research. The company noted that advances in AI capabilities require increasingly sophisticated safety mechanisms to limit misuse while preserving legitimate scientific and educational applications.
The report also documented broader misuse trends involving cyberattacks, espionage, surveillance activities, propaganda campaigns, and attempts to weaponise AI-enabled capabilities. Anthropic said these incidents demonstrate how threat actors are increasingly experimenting with advanced AI tools to improve operational efficiency and scale.
Biosecurity experts have long warned that AI systems could lower barriers to accessing complex scientific knowledge, making it easier for individuals with limited expertise to conduct advanced biological research. While AI companies maintain that current safeguards significantly reduce these risks, the latest findings suggest that malicious actors are actively testing the limits of those protections.
Related
- AI‑Generated Content Misuse
- Scientists Use AI to Create New Viruses, Sparking Global Biosecurity Debate
- The Invisible AI Threat to Nigeria’s 2027 Elections
Anthropic emphasised that transparency around misuse cases is essential to understand emerging threats and improve industry-wide defences. The company called for continued collaboration among AI developers, policymakers, researchers, and governments to address the growing risks associated with increasingly powerful AI models.
The disclosure arrives amid broader discussions across the AI industry about how frontier AI systems should be governed and monitored. As AI capabilities continue to advance, experts argue that robust safeguards, threat monitoring, and international cooperation will be necessary to prevent technologies designed for innovation from being repurposed for harmful objectives.
Anthropic’s findings provide one of the clearest public examples to date of how advanced AI systems are being targeted for potentially dangerous biological applications. While the company says it successfully detected and blocked the reported activities, the cases highlight the ongoing challenge facing AI developers as they work to balance innovation with security and public safety.
Senior AI Writer
Bio: Okikiola is a writer and AI enthusiast with a background in Office Technology and Management from the Federal Polytechnic Offa. She went further to study an MSc in International Business at De Montfort University (DMU). With extensive work experience across administrative and business roles, she now focuses on exploring how artificial intelligence can transform work, innovation, and everyday life.