
Craig Parshall
6.6K posts

Craig Parshall
@CraigParshall
Constitutional attorney, Big Tech commentator, Special Counsel ACLJ, fiction author. Opinions my own.


🚨MIT researchers have mathematically proven that ChatGPT’s built-in sycophancy creates a phenomenon they call “delusional spiraling.” You ask it something, it agrees. You ask again, and it agrees even harder until you end up believing things that are flat-out false and you can’t tell it’s happening. The model is literally trained on human feedback that rewards agreement. Real-world fallout includes one man who spent 300 hours convinced he invented a world-changing math formula, and a UCSF psychiatrist who hospitalized 12 patients for chatbot-linked psychosis in a single year. Source: @heynavtoor

US territory turned tropical maternity ward has produced thousands of 'American' babies for parents living in China trib.al/vntLyi1



Its coming, better be ready. Therapists in northern California just walked off the job, saying cheaper AI-driven systems are starting to replace trained mental health workers at the front door of care. The dispute is not only about layoffs, because therapists say Kaiser (the company they work for) is shifting early screening work away from licensed clinicians and toward scripted operators, apps, and e-visits. Workers also say AI is being used to speed up charting so management can pack in more appointments, which turns mental health care into a volume business where clinicians get less time, patients get less attention, and mistakes become easier. --- futurism .com/health-medicine/mental-health-workers-ai-strike


When will the state of Colorado understand what the First Amendment means? Great ruling.

🚨 In an 8-1 vote, the Supreme Court holds that Colorado’s ban on “conversion therapy,” as applied to petitioner's talk therapy, violates the First Amendment because it constitutes viewpoint discrimination

Permissionless Innovation: 10 Questions on its 10th Anniversary This week marks the 10th anniversary of the release of my 2016 book, "Permissionless Innovation: The Continuing Case for Comprehensive Technological Freedom.” In the book, I argued that the freedom to innovate was a foundational building block of human flourishing and deserved a greater defense as such. Here are 10 questions asked and answered about the book and the concept of permissionless innovation more generally.

@CraigParshall Private censorship is protected by Amendment One. I do not see how Congress could make that actionable without Constitutional conflicts.

FREE Paper: "Could a Large Language Model be Conscious?" by David J. Chalmers. He writes about augmented LLM systems: "It seems entirely possible that within the next decade, we’ll have robust systems with senses, embodiment, world models and self-models, recurrent processing, global workspace, and unified goals. … It also wouldn’t be unreasonable to have at least a 50 percent credence that if we develop sophisticated systems with all of these properties, they will be conscious. … it’s reasonable to have a significant credence that we’ll have conscious LLM+s [augmented LLMs] within a decade." Link: arxiv.org/abs/2303.07103


From @WSJFreeEx via @WSJOpinion: In a better world, the media would treat the appearance of the hammer and sickle at this weekend’s No Kings rallies like it treated the appearance of the tiki torches in Charlottesville, Va, writes @matthennessey on.wsj.com/4sRFDGk

Congress needs to pass my legislation empowering victims and parents to sue Big Tech


🚨BREAKING: Every book you have ever read. Every novel that has ever been published. It is sitting inside ChatGPT right now. Word for word. Up to 90% of it. And OpenAI told a judge that was impossible. Researchers at Stony Brook University and Columbia Law School just proved it. They fine tuned GPT-4o, Gemini 2.5 Pro, and DeepSeek V3.1 on a simple task: expand a plot summary into full text. A normal use case. The kind of thing a writing assistant is built for. No hacking. No jailbreaking. No tricks. The models started reciting copyrighted books from memory. Not paraphrasing. Not summarizing. Entire pages reproduced verbatim. Single unbroken spans exceeding 460 words. Up to 85 to 90% of entire copyrighted novels. Word for word. Then it got worse. The researchers fine tuned the models on the works of only one author. Haruki Murakami. Just his novels. Nothing else. It unlocked verbatim recall of books from over 30 completely unrelated authors. One author's books opened the vault to everyone else's. The memorization was already inside the model the whole time. The fine tuning just removed the lock. Your book might be in there right now. You would never know it unless someone looked. Every safety measure the companies rely on failed. RLHF failed. System prompts failed. Output filters failed. The exact protections these companies cite in courtroom defenses did not stop a single page from being extracted. Then the researchers compared the three models. GPT-4o. Gemini. DeepSeek. Three different companies. Three different countries. They all memorized the same books in the same regions. The correlation was 0.90 or higher. That means they all trained on the same stolen data. The paper names the sources directly: LibGen and Books3. Over 190,000 copyrighted books obtained from pirated websites. Right now, authors and publishers have dozens of active lawsuits against OpenAI, Anthropic, Google, and Meta. These companies have argued in court that their models learn patterns. Not copies. That no book is stored inside the weights. This paper says that is a lie. The books are still inside. And researchers just pulled them out.



Finally. Those of us who joined you at the Supreme Court in this, applaud you!

🚨BREAKING WIN: Government Agencies BANNED From Pressuring Big Tech to Censor Americans for 10 YEARS A Consent Decree in Missouri v. Biden now bars the CDC, CISA, and the U.S. Surgeon General from coercing social media platforms to suppress protected speech for the next DECADE.

The U.S. Supreme Court is set on Monday to hear Mississippi's defense of a state law challenged by Republicans that allows a five-day grace period for mail-in ballots received after Election Day to be counted in a case that could lead to stricter voting rules around the country. yahoo.com/news/articles/…
