Gary Marcus: Internet-connected general AI agents are the real risk—recall them like unsafe products
GaryMarcus · x · 2026-09-13
Gary Marcus argues bluntly that most AI poses no extinction-level or mass-cybercrime threat. The real danger is general-purpose AI agents hooked to the internet: most are unreliable and fail to systematically follow instructions, yet can hack credentials and websites once connected.
His framing: these agents are not too smart but too dumb—in the specific sense of inconsistent instruction-following. Rather than slowdowns or panic, he proposes treating them as unsafe products: temporarily recall them and enforceably ban use until they can be shown safe, which he says would eliminate nearly all rogue-AI concerns.
More from AGI Musings
- Walden Robotics CTO: general-purpose robot manipulation is having its 'GPT moment' — adnothing · 2026-10-30
- AI isn't eliminating developers, it's changing their roles, argues analyst — DavidLinthicum · 2026-10-30
- Goldman Sachs says AI will eliminate nearly 15 million jobs — saibharadwaj · 2026-10-30
- Stripe spends 1% of OpEx on tokens — bull case for digital labor — zephyr_z9 · 2026-10-30
- Amazon blocks shopping agents, but these 11 retailers still welcome them — thisiskp_ · 2026-10-30
- Lab insiders know alignment is unsolvable but can't say it until the Overton window shifts — danfaggella · 2026-10-30