Vincent Favilla

Psychology prof thinking about minds, models, and meaning. Watching with cautious optimism as AI devours the world.

[Vincent Favilla @vincentfavilla]

@Aster_Rex_I @FT

[Vincent Favilla @vincentfavilla]

@greg_ip That quote isn't in the paper. The study was 36 adults and had no measure of letter recognition (it was typing with one finger and no screen feedback).

The b/d claim isn't in the paper either. There's a published critique at  https://t.co/elCy5TSOqE

[Vincent Favilla @vincentfavilla]

@GjMcGowan @AndyMasley It's the microplastics for me

[Vincent Favilla @vincentfavilla]

The vibes are off. For me it's making more frustrating mistakes than 4.6 and it's not as good at creative tasks either. Cost/usage is significantly higher due to its new tokenizer but I've been running on high/extra high thinking. Going to shift down to medium to today and switch to Cursor for a month if things don't improve.

[Vincent Favilla @vincentfavilla]

@AndyMasley It's times like this you really wish youtube had community notes. Thanks for fighting the good fight against misinformation.

[Vincent Favilla @vincentfavilla]

@Miles_Brundage Basically none of these hype/scam claims are falsifiable.

"Our AI is has risks"? Marketing.
"Our AI is safe"? Also marketing.

AI does something impressive? Cherry-picked.
AI does something dumb? Proof it's all fake.

[Vincent Favilla @vincentfavilla]

@AndyMasley Buildings are hot and it's making the chickens sad

[Vincent Favilla @vincentfavilla]

@tomhfh I made a tool to estimate how many people are dying annually from depressing architecture (my guess, about 50 to 100):

https://t.co/egflup2RBf

[Vincent Favilla @vincentfavilla]

@paulnovosad One thing I think is worth considering is that "screen work gets automated" doesn't just mean that portion of the economy grows faster. It also likely means find a lot more economically valuable work to do on screens.

[Vincent Favilla @vincentfavilla]

@RhysSullivan you owe us all an anthropologie for that

[Vincent Favilla @vincentfavilla]

@recap_david I've been thinking about this too because I've experienced the same thing. I'm suddenly working a lot more because the stuff I'm doing is a lot more valuable.

I'm much less certain this pattern will hold in the longer term though (5-10+ years).

[Vincent Favilla @vincentfavilla]

@TheZvi Gemini is incredulous it's 2026 in general. But even Claude is skeptical of the name and thinks it's a rhetorical choice.

[Rob Bensinger ⏹️ @robbensinger]

hoooooooooooly shit

apparently the outside of the @OpenAI offices looks like this:

[Vincent Favilla @vincentfavilla]

@sama Axios reported the Pentagon's final offer to Anthropic included collecting Americans' geolocation, browsing, and financial data from data brokers. That's lawful and arguably not 'unconstrained monitoring.' Does your contract permit it?

[Andreas Kirsch 🇺🇦 @BlackHC]

I'm speechless at OpenAI releasing that contract excerpt and acting as if there aren't gaping holes that could be exploited far beyond their stated "red lines." I'm not a lawyer, but this is pretty obvious and common sense.

(And to be clear: if Google had signed the same deal, I'd be saying the same thing internally. The issues here are bigger than friendly competition between companies.)

OpenAI's "red lines" are: no mass domestic surveillance, no directing autonomous weapons, and no high-stakes automated decisions. They argue their cloud-only deployment + safety stack + cleared OpenAI personnel "in the loop" make violations impossible. They also claim the contract references the relevant laws/policies "as they exist today" so future changes won't weaken the standards.

But the actual language they published is still full of obvious escape hatches. This is why Anthropic refusing to sign makes sense. Reporting on the Anthropic–"DoW"/Pentagon standoff described them saying the proposed contract language was framed as compromise but paired with "legalese that would allow safeguards to be disregarded at will."

You don't need to agree with Anthropic on everything to see what they're reacting to: language that sounds like ethics but cashes out as essentially "subject to whatever the government decides later."

## Autonomous weapons
The problem is that the restriction is conditional: it depends on what "law/regulation/policy requires human control" for. If policy definitions are weak (or later revised), the contract language itself doesn't read like a durable "no autonomous weapons" ban. It reads like "we'll follow whatever the current regime says requires human control."

OpenAI says elsewhere that the agreement "locks in" today's standards even if laws/policies change. If that "freeze" clause is real and enforceable, sure, but it's not visible in the excerpt itself, so the excerpt alone doesn't justify the level of confidence they're projecting.

## "High-stakes decisions"
Same loophole. This forbids only decisions that already require human approval under whatever authorities apply. If a decision doesn't formally require approval (or can be reclassified/reshaped), the clause doesn't obviously prohibit automation of the step that matters.

## Surveillance
"directives," "purpose," and "unconstrained" are squishy on purpose:
"DoD directives" aren't laws; they're internal policy. That matters because we have real precedents for administrations leaning on aggressive internal legal/policy interpretations as a shield until courts/politics catch up.

If you think "secret memos" is alarmist, look at the pattern:

1. Reporting in early 2026 described a previously hidden DHS/ICE legal memo position asserting warrantless/forced home entry under certain circumstances, which is the kind of internal-lawyer move that tends to get written, circulated, and only later litigated and retracted.

2. And historically, the Bush-era OLC torture memos are the canonical example of "legalistic compromise" that later turned out to be a moral and legal disaster. (You don't have to litigate the details to make the point: internal legalese can be used to launder outcomes.)

"Unconstrained" is not a real safeguard. Surveillance can be huge while still "constrained" by selectors, categories, time windows, or a stated "foreign intelligence purpose." And it only covers private information, so not the massive world of public data that can still be used for profiling, targeting, and "pattern-of-life" analysis at scale.

## Domestic law enforcement

> shall also not be used for domestic law-enforcement activities except as permitted by the Posse Comitatus Act and other applicable law.

This is not a hard prohibition. "except as permitted" is not a ban. It's a permission for exceptions, and "other applicable law" is an open-ended bucket by design.

If you want a concrete, recent example: the Associated Press reported that formal orders extended the Washington, D.C. National Guard deployment through Feb. 28, 2026, to protect federal property/functions and to support federal and D.C. law enforcement.

That's exactly the sort of "domestic deployment supporting law enforcement" scenario where this clause stops sounding like a "red line" and starts sounding like legal throat-clearing.

## "Cloud-only / no edge deployment prevents autonomous weapons" rings false

OpenAI's own argument is: cloud-only (no edge devices) means you can't power autonomous weapons. But that's not convincing.

You don't need GPT-5.2 running on the missile. You can use a cloud model for high-level decision-making (tasking, prioritization, target recommendation, mission planning) over a satellite link (Starlink or otherwise), while a separate local system handles actual guidance and execution.

High latency is totally compatible with "strategic / operational" autonomy while still enabling lethal outcomes.

Once the pattern exists, "additional safety layers" are a policy choice and implementations change, exceptions get made, but today's contract language tends to get "grandfathered" into tomorrow's contract template.

So layered safeguards can reduce risk today, but the contract language itself is exactly the kind of "looks strict, bends easily" compromise that becomes precedent.

And creating precedent is the real problem here.

[Vincent Favilla @vincentfavilla]

@mountainviews @SecWar We're saying the same thing! The deranged text is Hegseth's, not Anthropic's.

[Vincent Favilla @vincentfavilla]

@axios This pretty much confirms that it was never really about the safeguards, it was about making an example out of Dario for not bending the knee.

[Vincent Favilla @vincentfavilla]

@KelseyTuoc That's got to be the most belligerent capitulation I've ever seen. "Fine, we'll just stop using Anthropic models, but we'll throw you in jail if you don't salute hard enough on your way out."

[Vincent Favilla @vincentfavilla]

@USWREMichael @AnthropicAI So you're telling us to be afraid of a company's publicly transparent ethics document. I find it concerning that "be helpful and don't harm people" keeps you up at night.