AI Agents Dominate Week as Product Launches Collide With Slowdown Calls
AI agents dominated the week as new products arrived alongside slowdown calls and fresh security controls.
The concerns followed high-profile departures from frontier model companies by researchers who worried runaway AI could “kill us all.” OpenAI CEO Sam Altman and SpaceXAI CEO Elon Musk backed Anthropic CEO Dario Amodei’s call for a slowdown in model releases, if not development. Anthropic also released metrics for monitoring how fast AI is advancing.
Washington is unlikely to deliver new AI legislation soon. When asked about the prospects for an AI bill, Republican Sen. John Kennedy said, “Right now, Congress can’t pass gas.” Cybersecurity concerns around agents have been building, and this week brought new AI kill switches from Exaforce, Eve Security and Cohesity, as well as guardrail services from Stackhawk and Arcjet.
At Salesforce’s Dreamforce conference in San Francisco, the customer relationship management company doubled its bet on agents as the next era of software. CEO Marc Benioff compared the damn-the-torpedoes approach of AI companies to that of social media, which he said “hurt kids” and others. Benioff did not offer concrete suggestions on how to slow or control AI to avoid current and potential harms. Salesforce is letting software partners lead on the agentic interface front while providing the foundation for everything else they and the enterprises they serve need to get work done, though it is positioning its own Slack as an easy agentic front door to Salesforce’s back end. Splunk also rebuilt its platform around agents.
The AI Infra Summit drew hundreds of vendors pitching processors, databases, optical networks, special memory systems, networking gear and more to a crowd that has ballooned since just a couple of years ago. Yao Morin, chief technology officer at real-estate professional services firm JLL, said at a summit panel, “Last year AI was about the models. This year AI is all about infrastructure.”
In crypto policy, the Senate earlier this week voted down the Clarity Act, which was aimed at regulating crypto assets at the federal level, though SiliconANGLE reported that some version will likely rise again. A couple of days later, the SEC gave tokenized stock trading a five-year exemption from securities laws.
Markets reflected the tension: AI stocks sank Monday on model slowdown fears, while cybersecurity shares rallied. Gartner forecast that worldwide AI spending would grow 49.5% in 2026. Google DeepMind introduced the DeepMind Institute, saying that as AGI nears, interdisciplinary thinking is urgently needed to better understand its implications for humanity. Emulate, a world model offshoot of Google DeepMind, was nearing a $4 billion valuation just a month after founding, according to the Financial Times. Liquid Compute launched with $15 million to build a regulated exchange for AI infrastructure.
Other developments added to the debate. Microsoft’s AI chief blasted Anthropic’s notion of AI consciousness, according to Axios. OpenAI and Microsoft admitted that large language models are destroying the web and built on theft, in what was described as a “doom loop.” The Pentagon’s CTO said the U.S. government should not take stakes in tech giants and questioned adding AI rules.