Anthropic warns of cyber danger from AI-exposed vulnerabilities
Updated
Updated · CNBC · May 5
Anthropic warns of cyber danger from AI-exposed vulnerabilities
7 articles · Updated · CNBC · May 5
Dario Amodei said Mythos has found tens of thousands of software flaws, with Chinese rivals possibly six to 12 months behind, giving governments, banks and tech firms limited time to patch them.
Anthropic has restricted Mythos to a few partners because undisclosed weaknesses could be exploited by criminals or hostile states, while Amodei and JPMorgan's Jamie Dimon said the risk could be temporary if addressed quickly.
At the same event, Anthropic expanded its financial-services push with 10 new AI agents and Microsoft Office integrations, while Amodei called for automotive-style AI guardrails that protect consumers without stifling competition.
Are AI tools training Wall Street's junior analysts for a new era of productivity or for their own eventual obsolescence?
As Wall Street adopts AI agents, who is liable when these autonomous systems inevitably make a costly financial error?
Can Anthropic's 'safe AI' approach win the high-stakes race for Wall Street against fierce competition from rivals like OpenAI?