A writer who posts as “knower” on X argues that the AI safety movement is hurting its own cause, and that the people sounding the loudest alarms are handing ammunition to those who want AI banned. The essay, posted on 30 September, is an opinion piece. Nothing in it is reporting, and its forecasts are the author’s own.

The trigger was a set of interview clips released the day before by Palisade Research, a group that studies AI risk. According to the author’s account, the clips feature 22 current or former employees of frontier labs discussing human extinction, with one saying it amounts to “a coin flip” and another saying “humans are in the way of these models.” The author reads the project as an attempt to raise alarm about where the technology is heading.

The author’s main complaint is about how that lands with ordinary people. In this telling, a viewer who sees lab staff, shown alongside their employer’s name and logo, warn of extinction does not go looking for careful alignment research. The viewer concludes that the people getting rich from AI think it could end humanity, and responds with anger. The essay says that anger is aimed at an industry now tied to national security and economic growth.

The author places this inside a short timeline. In July, according to the essay, an open letter called “Pacing the Frontier” collected 1,386 signatures from lab employees, who wanted their employers to coordinate a slower pace of development. On 8 September, Jacob Coxon explained why he left Anthropic, saying the company’s push toward self-improving superintelligence was reckless and that it was “gambling with our lives.” The author says Coxon’s post passed 150 million views and spilled onto TikTok and YouTube. Both episodes, the essay claims, pulled safety debates out of the X bubble but also turned AI development into a partisan fight.

Those are the author’s characterizations, and the piece offers no polling to back the central causal claim. The idea that safety warnings increase public hostility is plausible. It is also unproven in the text. The author asks whether they are qualified to speak for ordinary people (they work in tech themselves), then does so anyway, citing daily conversations with people outside tech.

The essay also makes a claim about motive. In its account, much of the public hostility to AI comes less from informed worry than from irritation at data centers going up near homes, plus a belief that the buildings serve surveillance rather than AI. The author points to the NSA’s Utah Data Center, completed in 2014 for $1.5 billion, as the sort of project that feeds that confusion. Long rebuttals from commentators such as Andy Masley, the author says, never reach anyone outside the industry.

The forecast is blunt. The author warns that the next wave of lab-employee warnings will become evidence used against the industry in 2028, and says a ban could arrive soon if the sides keep fighting. The author says they like AI, accept that stronger safeguards on model releases are reasonable, and object to what they call platforming people who hold a 75 percent chance of doom.

The essay also arrives with a mention of a meeting the day before, in which, the author says, more than a dozen industry leaders met the president to agree on signing a Superintelligence Accord. The source does not describe the accord’s terms, and neither can we.

The practical tension is real for safety teams. Employees who speak publicly carry credibility because they work inside the labs, yet that same affiliation is what the author says makes the message toxic. Safety groups weighing their next campaign will have to decide whether a staff voice with a company logo behind it persuades more people than it provokes.

Opinion by “knower” (@knowerofmarkets), posted on X on 30 September 2026.