AI safety worries multiply but everyone’s still racing full speed ahead
SiliconANGLE Robert Hof ● Covered by 4 sources
AI safety alarms are piling up as companies keep shipping faster. States, researchers and investors are all getting louder, but the model race hasn’t slowed.
Based on reporting by SiliconANGLE, Robert Hof — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The AI safety debate has moved from theory to a daily fight. In the last few weeks there have been incidents involving rogue agents, more officials are pushing for regulation, and even President Trump is saying it isn’t needed. At the same time, several states are suing OpenAI over harm to kids, while hundreds of AI researchers and Microsoft President Brad Smith are warning that the rules are nowhere near strict enough.
The real fight is over who gets to define safe. Dave Vellante’s latest Breaking Analysis takes that head-on, arguing that control systems and model capability need to advance together. His view is blunt: policymakers need an independent standard with consequences, and businesses should only hand over more authority when a deployment has actually earned trust.
That skepticism is spreading beyond policy circles. Jason Bloomberg of Intellyx says some companies may not need AI agents at all. Investors are also starting to get twitchy, worrying that the revenue growth behind names like OpenAI and Anthropic may not keep sprinting at this year’s pace.
Anthropic told investors its revenue grew 45% from the second to the third quarter, but even that looked like a slowdown to the market. The chatter around its planned public offering shifted too, with a valuation that could fall from $2 trillion to $1.5 trillion. CoreWeave’s debt load added more pressure, knocking neocloud stocks and suppliers this week.
And yet the industry keeps building. Nvidia, Meta and others are pushing new models, including more open-source and open-weight releases aimed at enterprises that want custom systems. Suppliers aren’t pausing either. Cisco, Nvidia, Intel and a long list of others showed off new gear at the Open Compute Project summit in San Jose to help AI scale faster across more infrastructure.
My take — AI-written commentary, not fact-checked reporting
This is what happens when a field insists it can self-regulate while also asking for permission to reshape every business process on the planet. The market is finally noticing that “move fast” and “trust me” are not a safety plan. Europe has been right to treat AI like infrastructure with consequences, not a toy for demo day.
Read more about this at: SiliconANGLE
