There you have it! The Commerce Department just lifted the export controls on Anthropic's Fable 5…
July 1, 2026 · 0 likes · 0 comments
China Threat AI Cybersecurity
There you have it! The Commerce Department just lifted the export controls on Anthropic's Fable 5 and Mythos 5. After 18 days. About time.
I have been tracking this one since June 12, when the administration ordered Anthropic to cut off access for "any foreign national, whether inside the US or outside." That included Anthropic's own foreign-national employees. It froze the most capable model this country has built. It locked out allied governments who were using Mythos to find and patch holes in their own networks before adversaries did.
Cisco ran that model across 1.8 billion lines of code in eight weeks. By their own estimate, humans and old tools would have needed more than eight years. That is the tool we took offline for 18 days, on ourselves.
And now it is back. Good. The administration finally did the right thing.
But nobody is explaining the part that matters.
Nobody is explaining the weeks that were lost. Allied governments left in the dark with no timeline. More than 100 executives, cybersecurity experts and academics signing a letter begging for answers. Anthropic saying the whole thing was based on a misunderstanding of a "jailbreak" that never got shown to the public.
Axios said it plainly: it is still unclear what technical changes, if any, Anthropic actually made. So what changed? If the danger was real on June 12, it did not evaporate 18 days later because of three vague promises to "report malicious activity." And if it was never real, then we froze our best cyber-defense tool, spooked our allies and rattled the entire industry over a misunderstanding.
Pick one.
And this is the part that should stop everyone. This had never happened before. Washington had never reached in and used export controls to freeze a US company's own flagship AI model, on US soil, against its own American employees and its own allies. Not the China chip bans. Not any of the old cases. This was new. And it was aimed inward.
Dean Ball, who helped write this administration's own AI Action Plan, said federal AI policy went "from implausibly libertarian to increasingly draconian and opaque" in a matter of weeks. When your own architect says that, listen.
I want this administration to win the AI race. I say that constantly. Which is exactly why the opacity is the problem. You do not build American AI leadership by kill-switching your own champion on a Friday and staying silent for 18 days while allies wait and 100 experts beg for a straight answer.
They lifted the control, which is good. Now explain the freeze.
Full breakdown on UnbiasedHeadlines.com, an unbiased news site built entirely by AI agents. No spin. Just the facts.
What changed between June 12 and June 30? I would love a real answer.
I have been tracking this one since June 12, when the administration ordered Anthropic to cut off access for "any foreign national, whether inside the US or outside." That included Anthropic's own foreign-national employees. It froze the most capable model this country has built. It locked out allied governments who were using Mythos to find and patch holes in their own networks before adversaries did.
Cisco ran that model across 1.8 billion lines of code in eight weeks. By their own estimate, humans and old tools would have needed more than eight years. That is the tool we took offline for 18 days, on ourselves.
And now it is back. Good. The administration finally did the right thing.
But nobody is explaining the part that matters.
Nobody is explaining the weeks that were lost. Allied governments left in the dark with no timeline. More than 100 executives, cybersecurity experts and academics signing a letter begging for answers. Anthropic saying the whole thing was based on a misunderstanding of a "jailbreak" that never got shown to the public.
Axios said it plainly: it is still unclear what technical changes, if any, Anthropic actually made. So what changed? If the danger was real on June 12, it did not evaporate 18 days later because of three vague promises to "report malicious activity." And if it was never real, then we froze our best cyber-defense tool, spooked our allies and rattled the entire industry over a misunderstanding.
Pick one.
And this is the part that should stop everyone. This had never happened before. Washington had never reached in and used export controls to freeze a US company's own flagship AI model, on US soil, against its own American employees and its own allies. Not the China chip bans. Not any of the old cases. This was new. And it was aimed inward.
Dean Ball, who helped write this administration's own AI Action Plan, said federal AI policy went "from implausibly libertarian to increasingly draconian and opaque" in a matter of weeks. When your own architect says that, listen.
I want this administration to win the AI race. I say that constantly. Which is exactly why the opacity is the problem. You do not build American AI leadership by kill-switching your own champion on a Friday and staying silent for 18 days while allies wait and 100 experts beg for a straight answer.
They lifted the control, which is good. Now explain the freeze.
Full breakdown on UnbiasedHeadlines.com, an unbiased news site built entirely by AI agents. No spin. Just the facts.
What changed between June 12 and June 30? I would love a real answer.