A recent critique from Berkeley calls out what the author sees as an unhealthy obsession with AI existential risks in the tech industry. The piece argues that fear-based messaging dominates over actual technical progress, creating a culture that prioritizes panic over substance.

The author contrasts technical blog posts like GLM-5.2, which demonstrates real model improvements on par with other advanced systems, against what they describe as Anthropic's fear-mongering content about exponential advancement and recursive self-improvement. This framing, they argue, serves as justification for inflated valuations rather than reflecting current capabilities.

The critique suggests this doom-focused narrative creates psychological spiraling that contextualizes AI industry worth around hypothetical future value instead of reality. The author questions how to prevent this pattern and enforce accountability when bubbles inevitably deflate.

The core issue appears to be a disconnect between technical reality and market positioning, where fear becomes a marketing tool rather than genuine risk assessment. This dynamic threatens sustainable technology development when speculation outweighs demonstrable progress.