Week in review #39
OpenAI, Google, Anthropic unveil agentic AI models.
New generation of agentic models
The industry’s leading AI developers announced a new generation of frontier models focused on “agentic capabilities.” The releases include GPT-6 Astra from OpenAI, Gemini 3.8 Flash from Google, Claude Fable 5.1 from Anthropic, and Grok 4.6 from xAI. These models are designed to enable autonomous interaction with software, conduct research, and execute complex, multi-step tasks, signaling a significant shift towards more autonomous AI systems for business automation.
US lawmakers push for strict AI oversight
US lawmakers introduced federal legislation aimed at strictly regulating advanced AI. The “Ban Artificial Superintelligence Act,” proposed by the Office of Senator Bernie Sanders and the Office of Representative Greg Casar, seeks to prohibit superintelligence and pause development until a new federal agency can establish safety rules. Separately, the California Governor's Office issued an executive order to explore the creation of an “AI kill switch” and accelerate independent safety verification for AI systems.
Massive investment in data center power
The immense energy requirements of AI data centers drove two major infrastructure deals. Amazon announced an $8 billion long-term agreement with Generac for backup power generators to support its expanding facilities. In a related move, Crusoe Data Centers, which specializes in high-density data centers for AI, closed a $3.9 billion funding round, tripling its valuation to $30.9 billion and underscoring strong investor confidence in the sector.
AI reshapes the graduate labor market
Recent studies indicate that AI adoption is having a measurable effect on entry-level professional roles. Research from the Federal Reserve Bank of Dallas found that recent graduates from AI-exposed majors at Texas universities saw a 1.7-percentage-point drop in their probability of employment and earned 5% lower wages since late 2022. Another study found that employment for 22-to-25-year-olds in AI-exposed fields is 19% below its expected growth trajectory.
Majority of US consumers adopt AI for shopping
AI-powered tools have reached a tipping point in mainstream consumer behavior. A report from NielsenIQ found that for the first time, a majority of US consumers, 51%, have used at least one AI-powered tool to support their shopping in the past month. AI-powered product recommendations were the most common application, used by 20% of shoppers, followed by personal shopping assistants at 16%.
Russian espionage group automates attacks with AI
A Russian state-affiliated espionage group has been observed using customized AI-driven workflows to automate and scale its cyber attack operations. According to a report from Anthropic, the group, identified as GTG-20006, used AI to automate phishing, malware development, and data exfiltration. The systems were also capable of monitoring their own effectiveness against security defenses and autonomously modifying malware until it could evade detection.