Trump drops restrictions on Anthropic's Mythos and Fable models 61%

By Tim Fernholz68%

6/30/2026, 7:16:06 PM

BS Summary: This article contains 26 faulty reasoning types, including Unattributed Quote, Appeal to Authority, and Hasty Generalization, with Biased Writer Voice as the most egregious example at 40.4% saturation with 169 hits. Analysis detected 1,288 faulty-reasoning hits from 418 analyzed words, generating a BS Score of 56.9% and a BS Rank of 61% (8,534 of 21,887 articles). This article is worse (more manipulative) than 61.00% of the article peer group.

The U.S. has lifted a requirement that Anthropic obtain a license before exporting its Mythos and Fable models abroad, a requirement that effectively cut off public access to what are widely considered the most advanced AI models released to date. 
The AI lab said it would begin restoring access to the models on Wednesday, July 1. 
On June 12, the U.S. government had added the products to its list of export-restricted technologies, meaning they could no longer be made available to foreign nationals without special approval. 
Complying with that rule proved impractical at scale, forcing Anthropic to end public access to the models altogether. 
Now, after weeks of talks, Secretary of Commerce Howard Lutnick said Anthropic “has agreed to proactively detect and address security risks associated with the models; to work diligently with the U.S. government on protocols and standards and releases for Mythos, Fable and future models; and to inform the US government of any malicious activity.” 
Anthropic had already publicly pledged to do much of this voluntarily, months before the export rule existed. 
That's part of why cybersecurity experts were skeptical of the restrictions in the first place. 
To them, the ban looked less like a security fix and more like leverage, a way for the Trump administration to punish Anthropic for its executives' public criticism of how the government, and the president's political opponents, might use the technology. 
Mythos was originally made available to a select group of organizations beginning in April to allay concerns about its ability to identify and exploit vulnerabilities in software, while a version called Fable was released to the public in June with additional security guardrails. 
However, with Asian AI companies beginning to release their own AI models approaching Mythos-level capabilities  among them Fugu and Tulongfeng  the U.S. government was under pressure to ease its restrictions on Anthropic to ensure that American AI could compete globally. 
Last week, Lutnick cleared Mythos to be released to select customers approved by the White House. 
OpenAI's latest models were also released to a group of organizations approved by the Trump team, instead of the public. 
The Trump administration's erratic approach to AI policymaking has left companies across the industry with little clarity about what will govern future model releases. 
An executive order issued in June that signaled a desire to review models ahead of release was criticized by influential analysts like Dean W. 
Ball, who recently started a policy position at OpenAI. 
Article reasoning-pattern comparisonThis article: 17.0%Tim Fernholz: 3.6%TechCrunch: 3.0%Confirmation Bias17.0%This article: 0.0%Tim Fernholz: 1.4%TechCrunch: 1.4%Anchoring Bias0.0%This article: 9.6%Tim Fernholz: 3.7%TechCrunch: 3.5%Availability Heuristic9.6%This article: 0.0%Tim Fernholz: 0.5%TechCrunch: 1.1%Representativeness Heuristic0.0%This article: 5.7%Tim Fernholz: 1.9%TechCrunch: 0.6%Hindsight Bias5.7%This article: 0.0%Tim Fernholz: 5.8%TechCrunch: 2.5%Overconfidence Bias0.0%This article: 19.6%Tim Fernholz: 3.9%TechCrunch: 4.8%Framing Effect19.6%This article: 0.0%Tim Fernholz: 0.4%TechCrunch: 0.6%Loss Aversion0.0%This article: 4.3%Tim Fernholz: 0.8%TechCrunch: 0.6%Status Quo Bias4.3%This article: 4.1%Tim Fernholz: 0.2%TechCrunch: 0.2%Sunk Cost Effect4.1%This article: 10.3%Tim Fernholz: 10.5%TechCrunch: 4.9%Optimism Bias10.3%This article: 0.0%Tim Fernholz: 0.9%TechCrunch: 1.3%Pessimism Bias0.0%This article: 19.1%Tim Fernholz: 2.9%TechCrunch: 5.0%Negativity Bias19.1%This article: 9.8%Tim Fernholz: 2.4%TechCrunch: 2.1%Self-Serving Bias9.8%This article: 2.2%Tim Fernholz: 1.2%TechCrunch: 0.5%Fundamental Attribution Error2.2%This article: 9.8%Tim Fernholz: 0.5%TechCrunch: 0.1%Actor-Observer Bias9.8%This article: 0.0%Tim Fernholz: 0.5%TechCrunch: 0.6%In-Group Bias0.0%This article: 0.0%Tim Fernholz: 0.5%TechCrunch: 0.3%Out-Group Homogeneity Bias0.0%This article: 0.0%Tim Fernholz: 8.8%TechCrunch: 3.5%Halo Effect0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.1%Horn Effect0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.0%Dunning-Kruger Effect0.0%This article: 13.9%Tim Fernholz: 1.9%TechCrunch: 2.3%Recency Bias13.9%This article: 5.7%Tim Fernholz: 0.8%TechCrunch: 0.3%Primacy Effect5.7%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.1%Blind-Spot Bias0.0%This article: 2.2%Tim Fernholz: 0.6%TechCrunch: 0.3%Ad Hominem2.2%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.6%Straw Man0.0%This article: 22.2%Tim Fernholz: 7.6%TechCrunch: 4.4%Appeal to Authority22.2%This article: 9.8%Tim Fernholz: 2.8%TechCrunch: 1.7%False Dilemma9.8%This article: 0.0%Tim Fernholz: 0.8%TechCrunch: 0.7%Slippery Slope0.0%This article: 0.0%Tim Fernholz: 0.6%TechCrunch: 0.2%Circular Reasoning0.0%This article: 20.1%Tim Fernholz: 3.7%TechCrunch: 6.0%Hasty Generalization20.1%This article: 0.0%Tim Fernholz: 0.2%TechCrunch: 0.2%Red Herring0.0%This article: 0.0%Tim Fernholz: 0.8%TechCrunch: 1.1%Bandwagon0.0%This article: 0.0%Tim Fernholz: 1.7%TechCrunch: 2.2%Appeal to Emotion0.0%This article: 4.1%Tim Fernholz: 0.3%TechCrunch: 0.6%Begging the Question4.1%This article: 14.4%Tim Fernholz: 2.4%TechCrunch: 2.9%Post Hoc (False Cause)14.4%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.1%Tu Quoque0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.5%Burden of Proof0.0%This article: 10.3%Tim Fernholz: 0.5%TechCrunch: 0.2%Appeal to Nature10.3%This article: 0.0%Tim Fernholz: 0.3%TechCrunch: 0.3%Composition/Division0.0%This article: 0.0%Tim Fernholz: 3.5%TechCrunch: 2.4%Anecdotal0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.1%No True Scotsman0.0%This article: 9.6%Tim Fernholz: 2.0%TechCrunch: 2.0%Ambiguity (Equivocation)9.6%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Tim Fernholz: 0.2%TechCrunch: 0.2%Middle Ground0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.0%Personal Incredulity0.0%This article: 0.0%Tim Fernholz: 0.0%TechCrunch: 0.1%Special Pleading0.0%This article: 0.0%Tim Fernholz: 0.1%TechCrunch: 0.1%Genetic Fallacy0.0%This article: 23.2%Tim Fernholz: 1.5%TechCrunch: 2.0%Unattributed Quote23.2%This article: 12.9%Tim Fernholz: 1.3%TechCrunch: 0.7%Quote-first Misdirection12.9%This article: 40.4%Tim Fernholz: 5.4%TechCrunch: 4.6%Biased Writer Voice40.4%This article: 2.2%Tim Fernholz: 0.1%TechCrunch: 0.8%Indoctrination2.2%This article: 0.0%Tim Fernholz: 1.6%TechCrunch: 0.1%Politically Left Leaning Bias0.0%This article: 5.7%Tim Fernholz: 0.3%TechCrunch: 0.1%Politically Right Leaning Bias5.7%This article: 0.0%Tim Fernholz: 2.5%TechCrunch: 4.9%Attempt to Sell a Product or S…0.0%

418 words analyzed.

Speakers

1speaker13%attributed speech364writer words
Voice mapSelect a segment to jump to its words
Writer's voice • 9 words • 100.0% coverageWriter's voice • 40 words • 100.0% coverageWriter's voice • 16 words • 100.0% coverageWriter's voice • 30 words • 0.0% coverageWriter's voice • 18 words • 100.0% coverageHoward Lutnick • 54 words • 100.0% coverageWriter's voice • 17 words • 100.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 41 words • 100.0% coverageWriter's voice • 43 words • 0.0% coverageWriter's voice • 42 words • 0.0% coverageWriter's voice • 16 words • 100.0% coverageWriter's voice • 20 words • 100.0% coverageWriter's voice • 24 words • 100.0% coverageWriter's voice • 24 words • 100.0% coverageWriter's voice • 9 words • 100.0% coverage
Selected voice

Howard Lutnick

100%flagged-word coverage
54 attributed words100% of attributed speech92% writer coverage
0%50.0%100.0%Quote-first Misdirection+100.0 ptsWriter: 0.0%Howard Lutnick: 100.0%100.0%Biased Writer Voice-46.4 ptsWriter: 46.4%Howard Lutnick: 0.0%0.0%Unattributed Quote-26.6 ptsWriter: 26.6%Howard Lutnick: 0.0%0.0%Politically Right Leaning -6.6 ptsWriter: 6.6%Howard Lutnick: 0.0%0.0%Indoctrination-2.5 ptsWriter: 2.5%Howard Lutnick: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.