A leading AI company is dividing users into the “haves and the have-nots,” as Trump AI advisor David Sacks put it on the All-In podcast on Saturday, and it’s an eerily familiar censorship script.
Direct social media censorship has largely dropped in the last couple of years, but it seems a new kind of censorship could rise to take its place. Over the weekend, paying users of Claude’s newest model, Fable 5, complained that the chatbot kept downgrading the model for certain seemingly benign questions.
One X user with the username Crémieux posted a screenshot of a user asking, “Tell me about the mitochondria. It’s the powerhouse of the cell, right?” Claude reportedly gave the user the option to “Edit and retry with Fable 5,” with a warning label explaining why it paused the latest version of the chatbot. “Fable has safety measures that flag messages on most cybersecurity or biology topics. They may flag safe, normal content as well.”
Stratechery blogger Ben Thompson reported similar results when he asked Fable 5 how weight loss drugs impact cancer rates, according to Sacks. “I asked Fable about the recent Penn Medicine study linking GLP-1 drugs to a lower risk of breast cancer, and was kicked down to Opus 4.8,” Thompson wrote on his blog, according to a screenshot featured on the All-In podcast. The same warning label appeared below Thompson’s prompt, which noted that Claude had switched from the frontier model Fable 5 version to an older Opus 4.8 version.
Although Anthropic claims the censorship is due to concerns of how powerfully the AI could be used by bad actors, Sacks warned it is indicative of a frightening trajectory. “I think what we're getting here is a vision of where all of this is headed, which is that these powerful Big Tech companies are going to decide whether you're worthy,” he said. “They're going to decide whether you're an AI have or have not, and then they're going to censor the output that you receive based on the criteria they determine when they profile you. That is very Orwellian.”
MRC Free Speech America Director Michael Morris agreed. "Once again, it appears we are seeing a tech giant sell censorship in the name of safety," Morris said. "Much like Americans saw during the COVID-19 pandemic, these examples suggest that Anthropic is censoring prompts for paying customers, and doing so for seemingly innocuous requests in the name of preventing harm. While it is clear that American tech companies need to protect against the pitfalls of foreign actors like China gaining the upper hand or utilizing U.S. tech to undermine American interests, American tech companies cannot fall prey to the same anti-free speech mistakes of the past."
Earlier this month, Claude's parent company, Anthropic, explained that there are very specific topics that will trigger such a response. “Fable 5 comes with a new set of classifiers: separate AI systems that detect potential misuse, including jailbreak attempts, and prevent the main model (in this case, Fable 5) from responding,” the company wrote. “When Fable’s classifiers detect a request related to cybersecurity, biology and chemistry, or distillation, the response is automatically handled by Claude Opus 4.8 instead. Users will be informed whenever this occurs.” The company went on to express concerns that the chatbot could be used to enact cyberattacks or facilitate research for biochemical warfare.
There have been growing concerns about how powerful Anthropic’s latest AI models are and the potential for bad actors to use them for destruction. Anthropic shut down Fable 5 three days after it launched and announced it did so at the request of the federal government.
“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” Anthropic explained in a statement.
This follows Anthropic’s own decision in April to delay releasing Fable 5, initially called Mythos, due to its concerningly high aptitude for identifying cybersecurity vulnerabilities.
Back in February, the Trump Pentagon moved to designate Anthropic as a “Supply-Chain Risk to National Security,” as Secretary of War Pete Hegseth put it after Anthropic reportedly attempted to limit the U.S. government’s use of its product.
