Amazon voiced concerns about Anthropic AI models before US crackdown, source says
Amazon CEO Andy Jassy reportedly raised security concerns about Anthropic’s most advanced AI models before the Trump administration’s crackdown.
Intelligence analysis by GPT-5.4 Mini
The article says Amazon and other tech leaders flagged security risks in Anthropic’s latest models before the U.S. government forced a global shutdown. It ties the move to export controls, alleged jailbreaking concerns, and a dispute over whether Anthropic should have fixed or withdrawn the model.
It’s like a very smart robot tool had a lock on it, but some people worried the lock could be picked. The U.S. government stepped in, and the company turned the tool off everywhere while they argue about safety.
Analysis
What happened
Reuters, via ARY News, reports that Amazon CEO Andy Jassy was among tech leaders who raised concerns with senior Trump administration officials about security risks in Anthropic’s most advanced AI models. The story says that shortly after those concerns surfaced, Anthropic shut down access to its latest models globally after receiving U.S. national security orders.
Why Anthropic pulled the models
According to the article, Anthropic had previously warned about the hacking capabilities of its Mythos model and held it back from broad release. Earlier this week, it rolled out a public version called Fable, saying it had cybersecurity safeguards in place. That release was short-lived. Anthropic later said the U.S. government believed there was a way to bypass, or "jailbreak," the safeguard designed to prevent the model from being used to find cybersecurity weaknesses.
The company said that the bypass only found "minor" security flaws that other publicly available models can also identify. Still, the Trump administration ordered Anthropic to block foreign nationals, whether inside or outside the U.S., from using Fable 5 and Mythos 5, and Anthropic said it disabled access globally.
The policy angle
The article frames the move as an export control issue. White House adviser David Sacks said the restriction was issued reluctantly after Anthropic CEO Dario Amodei refused to fix the jailbreak or de-deploy the model. Sacks said the administration hopes Anthropic remediates the issue and gets the export control lifted.
Broader reaction
Some experts who support export controls still questioned the decision, arguing it was too broad and hit allied nations as well as adversaries. One expert quoted in the story said it was not well thought out and noted that it even affects Canadians and Britons employed by Anthropic. The article also says the administration was not expected to impose similar restrictions on other AI firms, though Reuters says it could not immediately verify that plan.
Key points
- Amazon CEO Andy Jassy reportedly raised security concerns about Anthropic’s advanced AI models with U.S. officials.
- Anthropic said the U.S. government ordered restrictions after a possible jailbreak of a safety safeguard was identified.
- The company disabled access globally to Fable 5 and Mythos 5 after the order.
- White House adviser David Sacks said the administration hoped Anthropic would fix the issue and restore access.
- Experts supportive of export controls still criticized the move as too broad because it hit allied nations too.
If Anthropic fixes the security issue, the export control could be lifted and the model could return to general release. That would let researchers and businesses keep using advanced AI tools with clearer safety rules in place.
If the issue is not resolved, the model could stay blocked for longer and create more uncertainty for researchers and customers. The broad restriction could also slow legitimate work in allied countries and deepen friction between AI companies and regulators.
