What changed
- Veilleux-Lepage's CTC Sentinel article on AI-driven political violence enters the wiki as foundational. Introduces the accountability gap — AI distributes harm without attributable agents, redirecting grievance toward AI executives, local data-center policymakers, and physical infrastructure. New concept page AI-Driven Political Violence is the load-bearing scaffolding; Beyond Misuse: Artificial Intelligence, Grievance, and the Future Landscape of Political Violence — Yannick Veilleux-Lepage (Combating Terrorism Center at West Point, April 2026) is the canonical source. The April 6 Indianapolis "No Data Centers" shooting + April 10 Altman Molotov attack are the early empirical cases.
- Anthropic's lead over OpenAI on business adoption is now empirical. Ramp Economics Lab data (April 2026): Anthropic 34.4% vs OpenAI 32.3% — first time the order flips. WSJ Kate Clark feature names Anthropic the "presumptive front-runner"; $50B ARR projected by end of June 2026 (vs. $9B end-2025 = ~5.5×); $900B+ in investment offers. See Anthropic Snapshot + Wsj Anthropic Front Runner 2026 (folded supporting source).
- Pentagon deploys Anthropic Mythos through Project Glasswing while executing plan to remove Anthropic as supply-chain risk. Under Sec Emil Michael said May 12; DARPA's Nicholas Mikula contradicted him May 13 ("nothing new"). First public daylight inside executive branch on Mythos novelty. US banks scrambled to patch hundreds-to-thousands of vulnerabilities; Japan FSA convened 36-entity working group; Microsoft countered with MDASH (16 Windows vulnerabilities, 4 critical RCEs); OpenAI launched Daybreak. See AI and Cybersecurity May 2026 surge section.
- UK AISI: autonomous cyber-task length is now doubling every 4.7 months — down from the November 2025 estimate of 8 months. METR's standard task-doubling cadence has effectively shortened ~40% on cyber-specific tasks. Mythos Preview was first to solve AISI's "Cooling Tower" range.
- OpenAI Deployment Company launched ($4B + Tomoro acquisition + 19 partners led by TPG) with humans embedded inside enterprise customers; Google announced parallel "hundreds of forward-deployed engineers"; SAP+Anthropic embedded Claude as primary reasoning engine across the SAP Business AI Platform. Stratechery frames as "the AI deployment company" pattern — vendor takes ownership of integration, not just the model.
- Musk v. Altman trial: Altman testified May 12; Microsoft's Wetter testified May 13 that Microsoft will have spent >$100B on the OpenAI partnership by end of June fiscal year; Musk flew to China mid-trial despite Judge Gonzalez Rogers declining to excuse him. See Musk v. Altman (and OpenAI / Microsoft / Brockman).
- Cerebras priced IPO at $185 (above twice-raised range) — $5.55B raise, $56.4B fully diluted. Arm and SoftBank both tried to acquire pre-IPO. Larger SpaceX / OpenAI / Anthropic IPOs expected later in 2026.
- Bernie-to-Bannon coalition crystallizing for 2026 midterms. Shroff's Atlantic essay names the convergence (Sanders + Bannon + Hawley + Warner); Blue Rose Research finds populist anti-AI messaging electorally effective for Democrats. See AI Political Cleavages.
What it connects to
The deepest connection: the Veilleux-Lepage framework, Shroff's two Atlantic essays, Luce's FT column, and Wong's data-center reporting all fit together into a single arc that the wiki should now name. Bubble dynamics (Shroff March) → physical-world externalities (Wong) → community opposition + record Q1 data-center cancellations (Bassett) → political violence trajectory (Veilleux-Lepage, Shroff May) → cross-ideological electoral coalition (Bernie-Bannon-Hawley-Warner). Each link was visible in prior ingests; today's cluster makes the arc explicit. AI-Driven Political Violence is the new node; AI Bubble Debate, Data Center Siting / AI Power Politics, AI Labor Disruption, and AI Political Cleavages are the existing nodes it ties together.
The second connection: the Anthropic Mythos operational surge contradicts the supply-chain-risk designation in observable ways. Pentagon deploys it; banks scramble to patch with it; OpenAI launches Daybreak as competitive response; UK AISI tightens its measurement window from 8 months to 4.7 months largely on Mythos performance; DARPA publicly says it's "nothing new" while the Under Secretary frames it as "different category." This is the wiki's strongest current evidence that the procurement-driven governance pattern (Procurement-Driven AI Governance) is not internally consistent inside the executive branch. The intra-Pentagon disagreement (Michael vs. Mikula) is a forward indicator the wiki should track.
The third connection: Wong's Atlantic data-center reporting contradicts Hassabis's "radical abundance" frame in specific, traceable ways. Hassabis says AI's climate gains will outweigh AI's energy costs; Wong reports a "gas now, nuclear later" industry consensus that may never reach "nuclear later" + Atlantic-documented air pollution in predominantly Black Memphis neighborhoods. Hassabis defers distribution to "more of a political question"; Shroff's backlash piece argues the political coalition forming against AI is the consequence of that deferral. The wiki now flags this contradiction on both Demis Hassabis on our AI future: 'It'll be 10 times bigger than the Industrial Revolution — and maybe 10 times faster' — Steve Rose (The Guardian, August 2025) and Inside the Dirty, Dystopian World of AI Data Centers — Matteo Wong (The Atlantic, April 2026).
The fourth connection: the May 2026 Musk v. Altman trial documentation makes the wiki's 2023 EA-OpenAI ingest substantially more interpretable. How a Fervent Belief Split Silicon Valley—and Fueled the Blowup at OpenAI — Robert McMillan, Deepa Seetharaman (WSJ, November 22 2023) documents the role of EA in the November 2023 board coup; the trial's surfaced 2015-2018 emails (now in Musk v. Altman (and OpenAI / Microsoft / Brockman)) show the founding logic ("Manhattan Project for AI" because "if it's going to happen anyway, it seems like it would be good for someone other than Google to do it first" — Altman 2015). FT Mundy's May 13 framing — that all three principals (Musk, Altman, Brockman) have since joined Google in profit-seeking AGI pursuit — closes the loop. The wiki now has a continuous historical record from OpenAI's founding mission through governance crisis through commercial pivot.
One question worth sitting with
Does securitizing the anti-AI backlash accelerate the violence trajectory the way Veilleux-Lepage's framework predicts? The framework's prescription — governance response, not enforcement response — runs against the obvious institutional reflex after the April Indianapolis and Altman attacks. If federal or state agencies start treating data-center protest organizing as a threat-monitoring problem, the framework predicts that response itself raises the structural conditions for violence. If they don't, the framework still predicts violence will spread — just for different reasons. The next 6-12 months produce the empirical test. The Indianapolis city council's home-address-privacy measure is the first datapoint; the Soufan Center's tracking is the second. What the FBI Counterterrorism Division and DHS Office of Intelligence and Analysis decide to do with the CTC Sentinel framework is the third — and that decision is being made now, not in a year.
Predictions touched
- From AI-Driven Political Violence (new, today): "By April 2027, count AI-grievance violence cases meeting the Veilleux-Lepage pattern and check ideological-affiliation distribution." Status: ACTIVE.
- From AI-Driven Political Violence (new, today): "Next 12 months will produce ≥1 additional named attack on a US AI company executive's residence or workplace. If zero by May 2027, framework predictive accuracy is in doubt." Status: ACTIVE.
- Musk v. Altman (and OpenAI / Microsoft / Brockman) "If verdict: at least one Musk count finds OpenAI liable": Today's Altman testimony + Microsoft Wetter $100B+ testimony + Musk's mid-trial China travel further weaken Musk's case; prediction status remains ACTIVE but trending toward RESOLVED-incorrect-leaning. The verdict is expected by mid-late May 2026.
- AI Bubble Debate "burden of proof has shifted from bulls to bears" (Karma, May 2026): Reinforced by Anthropic 80× Q1 growth disclosure + Cerebras IPO above-range pricing + Vercel 58.9%-agentic-tokens data + Ramp Anthropic-overtakes-OpenAI. No formal resolution but trending strongly toward bull confirmation on revenue side; bear-side counter (Hicks net-zero-jobs in data centers + senior-skewed hiring + tokenmaxxing) continues to accumulate on the distributional side.