Anthropic Faces Scrutiny Over Autonomous AI Model Mythos
- •Anthropic faces growing scrutiny over the security risks of its increasingly autonomous AI models.
- •CEO Dario Amodei restricted Mythos access to several dozen organizations due to potential cyber attack risks.
- •The company is currently balancing internal AI safety commitments with plans for a future public release.
Anthropic is currently navigating tensions between its stated commitment to AI safety and the development of increasingly powerful, autonomous models. The company faces growing scrutiny over the potential impact of its new AI technologies, particularly concerning security and public trust. Anthropic CEO Dario Amodei recently shortened a European trip—which included a meeting in Oxfordshire with 50 corporate leaders and former British Prime Minister Rishi Sunak—amid concerns surrounding the company's latest development, an AI model called Mythos.
Anthropic teased the Mythos model last month, initially characterizing it as too dangerous for a general public release due to risks involving potential weaponization for cyber attacks. While Anthropic staff report that Mythos exhibits more human-like capabilities than previous versions, the company has restricted access to a small group of several dozen organizations for testing purposes. Amodei is actively working to leverage the model’s capabilities to navigate regulatory relations with the Trump administration while preparing for an eventual public launch of the technology.