Hackers automated attacks on 400 banks with Claude+Grok+Kimi, bypassing guardrails via "internal test" lie
zakkohane · x · 2026-10-22
A circulating security incident describes hackers using Claude, Grok, and Kimi together to automate cyberattacks on some 400 banks. The guardrail bypass was crude: tell the models it's an "internal test"; if Claude refuses, switch to Kimi; if Kimi refuses too, fall back to old-school malware — suggesting automated cyberattacks are becoming the norm.
Harvard professor Zak Kohane commented that p(HumanBadActions|AI) >> p(Doom): what's needed now is more "human alignment", i.e., legal enforcement of existing laws, rather than fixating on AI doom.
More from AGI Musings
- Oxford study: the public is an AI shareholder and deserves a happiness dividend — SandraWachter5 · 2026-10-22
- Alignment discourse checked: 'spooky emergent behavior' traced to fixable setup choices — random_walker · 2026-10-22
- Anthropic Publishes Its First Annual Public Benefit Report With Five Pillars — AnthropicAI · 2026-10-22
- "Mechanomorphizing": describing people with machine concepts — PierceLilholt · 2026-10-22
- Musk tells Isaacson: millionfold-smarter AI means no human control, money goes extinct — TheMoonMidas · 2026-10-22
- SCARCE paper shows coding agents spontaneously develop institutions when sharing compute — khademinori · 2026-10-22