Anthropic's Mythos uncovers tens of thousands of unreported software vulnerabilities
Anthropic's latest AI, Mythos, has uncovered tens of thousands of security flaws in critical software, including almost 300 in Firefox alone.
CEO Dario Amodei shared that these are far more vulnerabilities than previous AIs could spot, and most have not been made public yet to keep hackers at bay.
Anthropic warns, launches 10 finance agents
If these bugs are not fixed soon, they could open the door for big ransomware attacks on banks, schools, and hospitals (think serious financial trouble).
Amodei also pointed out that Chinese AI is catching up fast, so there is only a short window for US organizations to patch things up.
To help out, Anthropic has launched 10 new AI agents for finance and rolled out its top-performing Claude Opus 4.7 model into Microsoft Office to help automate complex tasks.