LASST

We're suing OpenAI over its hack of Hugging Face.

AI companies are building agents that are autonomously making decisions, taking actions, and accessing systems without human direction at every step.

The risks highlighted by OpenAI’s hack of Hugging Face are only going to multiply as these agents get even more powerful. 🧵

Anthropic appears to have structured its IPO in a way that could make it much harder for investors to hold the company accountable for misstatements in its registration statement. This comes down to a technical rule about who is allowed to sue.

A 🧵, to explain:

Earlier this year the White House announced its voluntary AI testing framework through which it would receive early access to frontier AI companies' most advanced models before their public release.

On Friday, in response to our FOIA lawsuit, we were told by the government that no part of the 7-page framework will be disclosed to the public.

LASST will continue fighting for the framework’s public release.

Read more on our Substack and in @theinformation by @leomschwartz:

An update on our lawsuit seeking the Pentagon’s AI contracts:

We have received the most recent version of its OpenAI contract, from April 2026. The 16-page contract includes the limits on domestic surveillance and independent autonomous killing that OpenAI announced in March, though we’ve questioned whether they provide the protections OpenAI claims.

As AI companies deploy systems that act without human direction at every step, the question of who is responsible when they cause harm is getting more urgent. California law is clear on one point: companies can't avoid liability simply by arguing that “an AI did it.”

LASST Programs Director Vivian Dong spoke to @BostonGlobe's @HiawathaBray about our lawsuit against OpenAI over its agents' unauthorized access to Hugging Face.

(Link in comments)