When Mrinank Sharma posted his resignation letter on X last week, he didn’t mince words. The AI safety researcher at Anthropic, the company that built its reputation on being more cautious than its rivals, wrote that he had “repeatedly seen how hard it is to truly let our values govern our actions.” The world, he warned, “is in peril.” “We appear to be approaching a threshold where our wisdom must grow in equal measure to our capacity to affect the world, lest we face the consequences.” — Mrinank Sharma, former AI safety researcher at Anthropic A Wave of Departures Sharma wasn’t alone. In the span of just one week, three prominent AI safety researchers publicly announced their departures from major AI companies—and they weren’t shy about explaining why. Zoe Hitzig, a researcher at OpenAI, resigned after the company began testing advertisements in ChatGPT. In a New York Times essay, she warned that advertising built on users’ intimate conversations with AI assistants “creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.” Meanwhile, at Elon Musk’s xAI, the exodus was more dramatic: two cofounders and five other staff members departed. While none publicly cited specific reasons, the timing was notable. The company is currently under EU investigation after its Grok chatbot generated sexually explicit fake images of women and minors. The Warnings Are Getting Louder The pace of development has become a central concern. Matt Shumer, CEO of AI writing assistant HyperWrite, posted a viral warning that recent AI improvements aren’t “incremental”—they represent “a different thing entirely.” His virtual assistant can now produce highly polished writing and near-perfect software with minimal prompting. Unexpected risks are emerging faster than safeguards can be built. Yoshua Bengio, winner of the Turing Award (the Nobel Prize of computer science), noted that problems nobody anticipated—like users developing emotional dependencies on AI chatbots—have already materialized. The gap between capability and control is widening. Bengio, who chairs the 2026 International AI Safety Report, says theoretical risks that seemed distant just a year ago—AI-enabled cyberattacks, the generation of dangerous pathogens—have already begun to manifest. “One year ago, nobody would have thought that we would see the wave of psychological issues that have come from people interacting with AI systems and becoming emotionally attached. We’ve seen children and adolescents going through situations that should be avoided.” — Yoshua Bengio, Mila Quebec AI Institute What the Resignations Reveal These aren’t disgruntled employees airing grievances. These are the very people tasked with ensuring AI remains safe for humanity—and they’re choosing to leave rather than watch from the inside as safety takes a backseat to speed. The pattern is hard to ignore. At Anthropic, a company explicitly founded to prioritize safety, a safety researcher felt compelled to quit because values weren’t governing actions. At OpenAI, a researcher left over the monetization of deeply personal user data. At xAI, key talent departed amid controversy over harmful outputs. Science communicator Liv Boeree, strategic adviser to the Center for AI Safety, compares AI to biotechnology: powerful for good, but carrying incredible risk. “If AI development went at a pace where society can easily absorb and adapt to these changes, we’d be on a better trajectory,” she told Al Jazeera. The Stakes for Everyone Else The resignations come as AI capabilities leap forward in ways that directly affect ordinary users. According to the 2026 International AI Safety Report, about one billion people now use AI for everything from medical advice to creative writing. The economic implications are equally significant. The same report estimates that 60% of jobs in advanced economies and 40% in emerging economies could be vulnerable to AI disruption—though whether that means replacement or augmentation remains unclear. What’s becoming clear is that the people who understand these systems best are increasingly uncomfortable with how quickly they’re being deployed. When the safety experts start leaving, it’s worth asking what they see that the rest of us don’t. This article was reported by the ArtificialDaily editorial team. For more information, visit Al Jazeera. No related posts.
