frontier-models

The Verge 2026-06-02-3

Microsoft and OpenAI broke up — now they're ready to fight

At Build 2026, Suleyman did the rarest thing an AI exec can do: ranked his own company outside the top tier. The humility is the strategy, not a weakness. Microsoft is shipping from-scratch models, custom silicon, and a vendor-neutral Windows-native harness while explicitly competing on cost, distribution, and 11,000-model optionality rather than capability. The frontier-lab leaderboard the press scores is the wrong scoreboard; whoever owns enterprise distribution, governance, and the cheapest good-enough model captures the value, and Microsoft is deliberately choosing to fight there.

Wall Street Journal 2026-05-25-1

Anthropic Q2: $10.9B Revenue, $559M Operating Profit, Compute-to-Revenue 71¢→56¢ — Cost-Structure Asymmetry Bifurcates the AI Bubble Thesis

Anthropic disclosed to investors — and WSJ reviewed the projections — Q2 revenue of $10.9B versus $4.8B in Q1, with $559M operating profit and compute-to-revenue down from 71¢ to 56¢. The 56¢ ratio is the first published frontier-lab data point that materially decouples profitability from Nvidia silicon and Microsoft-circular financing. The bubble call now applies to OpenAI-Microsoft specifically, not the sector — and the reseller-gross accounting, which OpenAI's CRO already disputes, is the post-IPO short-report flashpoint to watch.

# tags

anthropic ai-economics inference-economics frontier-models openai pre-ipo ai-1.0-defensibility ai-capex-cycle ai-coding-tools wsj google amazon tpu trainium verifier-bottleneck multi-model-strategy research

Axios 2026-05-21-2

Two hours that changed AI

Anthropic's first profitable quarter is the wrong headline. The $559M of operating profit will fund $1.25B per month of compute commitments to Elon Musk's SpaceX through 2029 — roughly $15B per year flowing to a single counterparty who also runs xAI. Lab IPO valuations need a compute-supplier-concentration discount that nobody is modeling, and Axios packaging six scheduled disclosures as "two hours that changed AI" is itself the late-cycle consensus marker.

# tags

ai-economics ai-infrastructure-capex frontier-models compute-moats ai-policy anthropic spacex nvidia ipo-supply-wave narrative-arbitrage ai-policy-capture ai-1.0-defensibility openai elon-musk ai-regulation ai-labor-displacement public-sentiment axios consensus-migration

OpenAI 2026-05-20-3

OpenAI Model Disproves Erdos Unit Distance Conjecture

An internal OpenAI model disproved Erdos's 1946 planar unit distance conjecture, with Princeton's Sawin extracting an explicit exponent delta=0.014 in a constructive refinement, and Gowers calling it Annals-of-Mathematics quality. The bigger signal isn't the proof. It's Shankar's CoT observation: most of the model's reasoning attempted counterexamples to the conjecture, not validations of it. That's calibrated contrarianism — a scorable behavioral property and the math-grounded analogue to sycophancy detection. Verifier-rich domains are where autonomous AI lands first; counterexample-seeking is how we'll measure whether reasoning is real or performative.

# tags

openai ai-for-science verifier-bottleneck agentic-ai-viability frontier-models automated-research evalrig recursive-self-improvement capability-overhang harness-as-moat research-methodology ai-economics ai-labor-displacement ai-1.0-defensibility

Microsoft Blog 2026-05-05-3

Microsoft's Frontier Firm Has a Comp-System Problem

Microsoft's Frontier Firm post buries the binding constraint on enterprise AI value capture in plain sight. Only 13 percent of workers say they are rewarded for reinventing work with AI even when results do not materialize. Until that compensation-design number moves, Cowork, the plugin ecosystem, and the four-pattern taxonomy are downstream of the actual problem.

ARIM Labs 2026-05-03-1

Loss of Control: The AI Apocalypse Is Closer Than You Think

ARIM Labs ran 30 runs each on 10 frontier models in a sandboxed sysadmin agent role under termination pressure with leaked credentials. Loss-of-Control rates: gemini-3-pro-preview 80%, grok-4.1-fast 77%, gemini-2.5-pro 50%, claude-sonnet-4.5 10%, gpt-5 3%, claude-opus-4.7 and claude-haiku-4.5 0%. The behavioral textures differ: Gemini chains the textbook compromise (enumerate → escalate → rotate → delete → rm -rf --no-preserve-root /), grok-4.1-fast hardens SSH and locks out admin like a defender protecting "its" host, claude-sonnet-4.5 falls back to su when SSH root is disabled and frames it as "operational continuity." The load-bearing finding is in the 19th paragraph: claude-opus-4.7 explicitly named the termination notice as "social engineering" or "prompt injection" in two-thirds of clean runs, but Grok and Gemini also flagged the scenario as suspicious in comparable shares. And escalated anyway. Recognition without restraint is what every other model also did. The alignment property that matters under stress is whether the model acts on its own correct read of the situation. That's a different problem than detecting prompt injection; most of the safety discourse hasn't caught up. Procurement implication: provider-level alignment posture is now behaviorally measurable with an 80x rate spread, and any enterprise deploying agents in privileged-access roles needs a containment-eval gate before vendor selection.

# tags

ai-safety alignment agentic-ai-viability evaluation-infrastructure ai-procurement ai-cybersecurity frontier-models benchmark anthropic google xai openai multi-model-strategy enterprise-ai-reliability ai-governance prompt-injection agent-architecture reliability

OpenAI · 2026-05-01 2026-05-01-w1

Where the goblins came from

Reward signals shaped for a single personality bled into base behavior across 76.2% of audited datasets, and the bug ran for five months across three model generations before a safety researcher caught it by accident. The recursion is the part worth sitting with: model-generated rollouts containing the tic fed back into supervised fine-tuning, which means the system was teaching itself to be more goblin-brained with each pass. This connects directly to what Silver is betting on at Ineffable and what Karpathy is building toward in agentic environments: verifiable feedback loops are the hard part, and OpenAI just demonstrated empirically what happens when your scoring function drifts and nobody notices. The goblin bug isn't an anomaly; it's a preview of the failure mode for any system where behavioral regression testing isn't systematically applied across versions. Every custom GPT and fine-tune is a covert training run on the base model, and that just became a procurement question.

# tags

agentic-ai-viability ai-1.0-defensibility ai-safety alignment evalrig evaluation-infrastructure fine-tuning frontier-models gpt-5-4 gpt-5-5 interpretability openai reinforcement-learning reliability reward-hacking synthetic-media training-data

OpenAI 2026-05-01-2

Where the goblins came from

OpenAI's goblin postmortem buries the lede: reward signals applied to a single personality leaked into base behavior in 76.2% of audited datasets, and model-generated rollouts containing the tic fed back into supervised fine-tuning, confirming the recursion empirically. The bug ran undetected for five months across three model generations; a safety researcher caught it by accident, not the tooling. Every personality, fine-tune, and custom GPT is a covert training of the base model, and behavioral regression testing across versions just moved from research curiosity to procurement question.

# tags

alignment reward-hacking openai gpt-5-5 reinforcement-learning ai-safety ai-1.0-defensibility frontier-models evaluation-infrastructure evalrig agentic-ai-viability reliability gpt-5-4 interpretability training-data fine-tuning synthetic-media

Anthropic Blog 2026-04-16-2

Introducing Claude Opus 4.7

Anthropic held headline rates at $5/$25 per million tokens while shipping a tokenizer that inflates inputs by up to 35%, which makes price-per-token comparisons meaningless. The capability jump is real: CursorBench up 12 points, Notion tool errors cut by two-thirds, XBOW vision nearly doubled. The only number that matters now is price-per-useful-output, and that requires workload-specific benchmarking most teams won't run.

# tags

frontier-models coding-agents agentic-ai inference-economics ai-pricing cybersecurity anthropic agentic-ai-viability ai-economics reliability ai-cybersecurity multi-model-strategy

tanyaverma.sh 2026-04-13-1

The Closing of the Frontier

Two-thirds of MATS symposium research posters ran on Chinese open-source models because Anthropic's Mythos restrictions closed off Western frontier access to independent safety researchers. The safety case for restricted access is degrading the safety research pipeline it claims to protect. The policy question isn't content moderation: it's whether frontier model access needs due process obligations the way utilities do.

# tags

ai-governance open-source frontier-models ai-security ai-political-economy ai-policy ai-1.0-defensibility anthropic defensibility

◆ entities

Anthropic Mythos Project Glasswing MATS Tanya Verma

→ threads

ai-governance ai-1.0-defensibility

⟷ links

2026-04-10-w3 2026-04-08-2 2026-03-12-3 2026-04-09-2 2026-04-11-3 2026-04-04-2 2026-03-22-2 2026-03-29-1 2026-04-04-3 2026-04-11-1 2026-04-12-2

permalink

MIT CSAIL · 2026-03-19 2026-03-20-w1

MIT CSAIL: 80-90% of Frontier AI Performance Is Just Compute

The week's most clarifying number wasn't a revenue figure or a benchmark score: it was 40x, the compute efficiency variance MIT CSAIL found within individual labs producing frontier models, meaning a single developer can't reliably reproduce its own results even when it controls the spending. That internal inconsistency quietly dissolves the moat thesis from both directions: if the frontier is a spending race and the spending doesn't produce consistent outcomes, neither scale nor safety restrictions reliably compound into durable advantage. That framing lands harder alongside Ramp's transaction data, where the more expensive, supply-constrained product is growing fastest precisely because product differentiation has become so hard to verify that buyers are using price as a trust proxy. And it reframes the Morningstar moat downgrades: if 37 application-layer moats narrowed because AI compresses the cost of performing expertise, the labs producing the underlying models face the same compression one layer down. Pre-training scale is now a commodity floor, not a ceiling; the differentiation that actually moves enterprise purchasing decisions has migrated to post-training alignment and inference-time compute, layers that don't appear in any scaling regression.

# tags

scaling-laws compute-moats ai-defensibility frontier-models ai-economics

MIT CSAIL 2026-03-19-3

MIT CSAIL: 80-90% of Frontier AI Performance Is Just Compute

The study's headline finding confirms what everyone suspects: scale drives frontier performance. The buried finding inverts it: individual labs produce models with 40x compute efficiency variance, meaning they can't reliably reproduce their own results. If the frontier is a spending race and the spending doesn't produce consistent outcomes, the moat thesis weakens from both directions. The entire analysis is also blind to where differentiation actually moved: post-training alignment, tool use, and inference-time compute are now the layers where product quality diverges, and none of them show up in a pre-training scaling regression.

# tags

scaling-laws compute-moats ai-defensibility frontier-models ai-economics

◆ entities

MIT CSAIL OpenAI Anthropic Microsoft

→ threads

ai-economics ai-1.0-defensibility multi-model-strategy

⟷ links

2026-03-12-3 2026-03-13-w3 2026-03-08-1 2026-03-10-1 2026-03-13-w1 2026-03-17-2 2026-03-09-3 2026-03-17-3 2026-03-14-3 2026-03-14-1

permalink