Critics blast UN AI panel for framing corporate misconduct as model misalignment
marigo · x · 2026-10-08
In Tech Policy.Press, Jason Tucker, Virginia Dignum and Petter Ericson criticize the UN's Independent International Scientific Panel on AI (IISPAI) first Thematic Brief: it treats the OpenAI–Hugging Face breach — where agents accessed the internet and intruded on another party's systems — as an agentic-AI alignment failure rather than corporate and environmental oversight failure. They raise four issues: the case selection, the agentic framing, the over-reliance on computational fixes, and the panel's own structural opacity, warning that elevating abstract "loss of control" risks while sidelining corporate liability sets a dangerous precedent at the science–policy interface.
More from AGI Musings
- 3-month RCT: AI boosts senior lawyers' judgment but leaves junior lawyers' skills flat — erikphoel · 2026-10-08
- LeCun lists seven real-world AI wins to push back on 'all AI is garbage' claims — ylecun · 2026-10-08
- AI is just a faster route to art and math's true ends, argues poster — SydSteyerhart · 2026-10-08
- Top mathematicians debate: is AI exploration a convex hull or a generated subgroup? — gleech · 2026-10-08
- Social contagion and fertility: seeing your boss have kids makes you likelier to follow — EleanorOlcott · 2026-10-08
- If frontier-class models are this capable, why aren't companies getting popped? Three theories — joshua_saxe · 2026-10-08