Anthropic safety researcher quits, warning 'world is in peril' (www.semafor.com)

A safety researcher from Anthropic, Mrinank Sharma, has resigned, expressing alarm over the rapid advancements in AI and asserting that the "world is in peril." Sharma highlighted the internal pressures within the safety team to prioritize productivity over essential safety concerns, including the risks of bioterrorism. Anthropic, established with a mission to develop safe AI, faces increasing scrutiny from its own team members about the pace of AI development and its potential dangers. CEO Dario Amodei has echoed similar sentiments, advocating for regulatory measures to slow down AI progress at industry forums like Davos. This resignation underscores a growing trend within the AI community, where safety researchers are becoming increasingly vocal about the catastrophic risks associated with unchecked AI advancement. The recent departures from Anthropic and key teams at OpenAI signal a potential crisis in the sector, highlighting a tension between financial incentives and the ethical responsibility to minimize risks inherent in creating highly intelligent AI systems. These events point to a need for stronger governance and ethical frameworks in AI development to ensure that safety does not take a backseat to rapid innovation.