GOP Targets Altman’s OpenAI Deals—SEC Probe Looms for AI’s Power Players
Republican AGs push SEC to scrutinize Sam Altman’s business entanglements—why builders should track the fallout. Plus: AI job forecasts clash, and tools for agentic workflows.
The story of the day
Republican state attorneys general have formally asked the SEC to review Sam Altman’s business dealings, escalating scrutiny over potential conflicts of interest at OpenAI. The request—filed by AGs from Tennessee, Alaska, and Nebraska—cites concerns over Altman’s overlapping roles across OpenAI, its investment funds, and affiliated startups, including the $2B "Conflict Map" we covered last week. For builders, this isn’t just regulatory theater: OpenAI’s API stability and model roadmap now hinge on governance outcomes. If the SEC acts, expect delayed releases, stricter compliance burdens for partners, and a possible exodus of enterprise customers hedging against instability. The irony? Altman’s empire was built on avoiding the very bureaucracy now threatening it.
New tools & launches
Agent Thread — Share Claude/Codex sessions as public, threaded links for async collaboration. Open-source, MIT-licensed. [HN]
Auto Agent Protocol — Vertical for AI agents buying/selling cars (yes, really). Ethereum-based, targeting dealership automation. [Site]
Shitty AI App Idea Generator — Satirical take on the AI gold rush. Because sometimes the signal is in the noise. [HN]
Models & research
Gartner’s AI job forecast flips the script: net job creation from AI starts in 2028, with 1.8M new roles annually by 2030. The catch? 72% will require "AI literacy"—a skill most devs still lack. Contrast this with King’s College London’s study (also out today), where 1 in 5 Britons fear AI-driven civil unrest by 2035. The gap between analyst optimism and public sentiment has never been wider.
Zknill’s AI application taxonomy proposes three generations:
- Conversational (2020–2023: chatbots)
- Delegative (2024–2026: agents handling tasks)
- Collaborative (2027+: humans and AI co-creating).
Builders note: The shift from #2 to #3 demands new trust primitives—expect a surge in agent auditing tools.
Industry pulse
On expired domains and AI agents: DomainAlot’s analysis reveals legacy SEO metrics (DA, backlinks) fail when AI agents—unlike humans—ignore domain history. The takeaway? "Brand authority" now depends on real-time utility, not decaying signals. [Substack]
AI quant desks emerge: Raster Finance’s onchain portfolio analytics tool lets traders backtest strategies using LLM-generated hypotheses. The twist? It’s not about replacing quants—it’s about augmenting them with 10x more simulations. [Site]
Stat of the day
1 in 5 Britons believe AI will cause civil unrest by 2035—higher than fears of climate change (17%) or economic collapse (15%). Source: King’s College London, May 2026.
Get the full analysis—subscribe to Gradient Pro.
