Technology

US forces Anthropic pull models amid bypass claims

US forces – As the week was ending, the US government ordered Anthropic to remove its two newest models, Fable 5 and Mythos 5, citing national security concerns after Amazon researchers allegedly demonstrated a way to bypass Fable 5’s guardrails. Cybersecurity researchers

Just as last week was ending, the US government forced Anthropic to pull its two newest models, Fable 5 and Mythos 5. The reason was national security concerns tied to an allegation that Amazon researchers found a way to bypass Fable 5’s guardrails.

It didn’t take long for the pushback to appear. Cybersecurity researchers have since signed an open letter calling the government’s move dangerous. In parallel. Anthropic itself said the jailbreaks being discussed weren’t unique to those specific releases. noting that the same jailbreaks exist in other models.

That tension—between a sudden government order and a claim that the underlying weakness already exists elsewhere—has quickly become the story’s fault line. Developers who build on Anthropic’s platform are left to wonder what this means in practice: if the issue stretches beyond the models that were pulled. what exactly does removal change?.

The controversy is also landing in the middle of broader market expectations around Anthropic’s IPO, with the moment framed as part of a complicated relationship between the company and the Trump administration.

On the episode of TechCrunch’s Equity podcast. hosts Anthony Ha. Sean O’Kane. and Rebecca Bellan unpack what the ban means for developers and for anyone watching the IPO. They also discuss why the move might. in a way that’s hard to reconcile with the stated national security concerns. be “accidentally” good for the company—alongside the week’s other headlines.

Whether this is a genuine security intervention or the latest messy turn in a fraught political and technical standoff. the sequence is now clear: government pressure. external research pushback through an open letter. and Anthropic pointing to the same jailbreaks in other models. The gap between those pieces is where the uncertainty has started to compound.

Anthropic Fable 5 Mythos 5 US government ban national security concerns guardrails bypass Amazon researchers cybersecurity open letter jailbreaks IPO Equity podcast TechCrunch generative AI safety

4 Comments

  1. I don’t even get how “national security” works here. If Amazon found a bypass then doesn’t that mean everyone already has it?? Seems like politics more than security.

  2. This is kinda like they pulled the wrong model? Like if the guardrails are bypassable on Fable 5, removing it won’t stop jailbreaks on Mythos 5 or whatever. Also the article mentions IPO expectations so maybe this is just to boost prices later.

  3. Open letter this, guardrails that… meanwhile regular people just want it to work. Doesn’t make sense that Anthropic says it’s not unique while the government is acting like it is. Feels like someone’s trying to cover their butt after rumors from “Amazon researchers” or something. Also Trump admin?? Ugh.

Leave a Reply

Your email address will not be published. Required fields are marked *

Are you human? Please solve:Captcha


Secret Link