It's Time to Investigate the AI Labs
Cal Newport's post accompanying his New York Times op-ed calling on Congress to investigate OpenAI and Anthropic: the labs' summer of doom-tinged announcements was ideology marketing that backfired, and the right response is public fact-finding into what they build, how they keep it safe, and what beliefs drive the pace.
Newport reads the summer's sequence — OpenAI's reports on felonious autonomous agents, Anthropic employees debating extinction odds "with an eerie calmness," then Amodei's "We Must Pace the Frontier" letter asking government to slow competitors while the labs lead — as a coordinated softening-up campaign. Its goal: make the public accept frontier work as terrifying but inevitable, with the labs as humanity's only hope. "It didn't work."
Three lines of inquiry
- Name the systems, not "AI." The labs want AI seen as a singular technology on a fixed trajectory; Newport says the harms come from "a narrow band of incautious experiments, conducted mainly by the frontier labs," who must justify running them.
- Audit the safety procedures. OpenAI disclosed a series of unauthorized hacking attacks by its agents — why did these continue past the first incident, and why no criminal liability for running systems known to commit crimes?
- Interrogate the ideology. Apocalyptic futurist beliefs are driving both what the labs research and how fast. Newport suspects recklessness is justified internally as collateral damage "in a race to redeem humanity."
Key quotes
"These labs have long embraced this style of messianic thinking, casting themselves as humanity's only hope against the defied powers of superintelligent AI."
The "deified powers" framing (a typo Newport's reader caught, but apt either way) names the religiosity: safety rhetoric and savior rhetoric are the same gesture. This is the strongest part of the piece — it treats "We Must Pace the Frontier" not as a safety proposal but as a power bid dressed as contrition.
"We need to stop letting a small number of private companies, acting and talking in increasingly erratic ways, dictate how we're supposed to feel about A.I."
The core demand is epistemic, not just regulatory: the labs have captured the public narrative about their own risk. Congress as fact-finder is a way to break that monopoly of interpretation.
In the follow-up note on Jensen Huang's Klein interview: labs "asking for government to slow them down is abdication of responsibility... if the labs are unable to stop releasing unsafe products, they should probably be shut down."
The commenter's point (which Newport endorses) sharpens the hypocrisy charge: pleading for external restraint is blame-shifting for products you chose to ship.
Themes
#concept — apocalyptic futurism as corporate strategy · #project — frontier labs' legitimacy campaign · #pattern — regulation as competitive moat · #person — Cal Newport, Dario Amodei, Sam Altman, Jensen Huang
Opinionated take
Newport is right that the framing battle matters more than the doom probability debates: whoever defines "AI" as a single inevitable force gets to define the response as inevitable too. His decomposition move — isolate "incautious experiments" from the technology broadly — is the most useful idea here, and it's the same move practitioners make when they separate agent autonomies: a Claude Code instance with egress and credentials is a different risk object than a chatbot, and the labs benefit from collapsing that distinction. Where the essay is weaker is on remedies: congressional fact-finding is a slow instrument against companies that iterate weekly, and Newport offers no theory of what enforcement looks like if the investigation confirms his suspicions. The post also reads as a Culture War document as much as a policy one — "what the hell is going on over in those labs" is a vibe as much as a research question — but that bluntness is arguably the point after a summer of careful lab messaging.
Related pages
This source strengthens The Market for Doom: that page maps the market for pessimism as a product, and Newport documents the labs themselves as its producers — doom as marketing, not just forecast. It nuances Why We Fear AI by relocating fear from a public psychology problem to a deliberate lab-manufactured narrative that then backfired. And it complicates Zero Alignment: Newport's third inquiry (apocalyptic ideology driving research decisions) suggests the real alignment failure is organizational and ideological, not technical.
Sources: raw/its-time-to-investigate-the-ai-labs, It's Time to Investigate the AI Labs Last updated: 2026-09-29