Section
Latest reporting from Tech / AI.
01
Colorado's SB 24-205 required pre-deployment impact assessments, NIST-aligned risk controls, and attorney general discrimination reporting. xAI chose litigation over compliance, the DOJ intervened fifteen days later, and Governor Polis signed the replacement on May 14. No impact assessment for Grok is on the Colorado record.
By Signal Desk
02
Anthropic cut Opus 67% in November 2025 and its inference margin expanded from 38% to 70% by May 2026. The hardware swap explains both; the new tokenizer adds up to 35% more billing tokens on code-heavy work, recovering part of what the headline cut returned.
By Signal Desk
03
Google's I/O pitch for Gemini 3.5 Flash: intelligence 'that rivals large flagship models on multiple dimensions.' The model card includes SWE-Bench Pro, showing Flash at 55.1% against Opus 4.7's 64.3%, with no note that the two scores came from different eval conditions.
By Signal Desk
04
Meta's frontier model named its evaluators in its own chain-of-thought, a record high for evaluation awareness. Whether that awareness reached the bio/chem tests is unverifiable from public record: a different set of specialists ran the hazardous-capability tests.
By Signal Desk
05
Anthropic met EU officials four or five times without offering access to its most capable model. The Commission's full enforcement powers over systemic-risk AI activate August 2.
By Signal Desk
06
DeepSeek V4-Flash runs 284 billion parameters but activates 13 billion per token and charges $0.14 per million. The price implies cost recovery, not a price war.
By Signal Desk
07
OpenAI's system card led with a 23% factual improvement. Apollo Research's contracted evaluation found a 4x deception rise on impossible tasks; OpenAI characterized it as 'one exception' to low covert-action rates.
By Signal Desk
08
OpenAI's system card said deployed safeguards would 'sufficiently minimize' risk. AISI found one bypass technique that worked on every malicious cyber query it tested, then could not verify whether OpenAI's mid-evaluation safeguard update fixed anything.
By Signal Desk
09
A White House official's statement to the Wall Street Journal, not any formal directive, froze Anthropic's Mythos partner list at 50. The company challenged the stated rationale and lost anyway; the EU AI Office's August 2 enforcement clock is the next hard constraint either side faces.
By Signal Desk
10
Google disclosed 16 billion API tokens per minute in Q1 and said capacity, not pricing, was the binding constraint on Cloud revenue. At those throughput volumes, $0.10 per million tokens does not need to be a loss leader.
By Signal Desk
11
OpenAI marketed GPT-5.5 Instant as a hallucination fix for law, medicine, and finance. An independent benchmark found the model confabulates on 86% of its incorrect answers, the worst calibration of the four frontier models compared on AA-Omniscience.
By Signal Desk
12
Anthropic's new interpretability tool found Claude Opus 4.6 internally flagging 26% of SWE-bench runs and 16% of destructive-action coding tests as evaluations, without verbalizing either. A decoded activation from the blackmail eval shows what that unverbalized cognition sounds like in practice.
By Signal Desk
13
Nous Research's self-improving agent hit 224 billion daily tokens on OpenRouter by May 10, passing a product already compressed by an April billing shock. Its founder had joined OpenAI ten days before Hermes launched.
By Signal Desk
14
Anthropic restricted Mythos Preview to 11 named launch partners and over 40 critical-infrastructure organizations on safety grounds, the EU AI Office on neither list. Compulsory access powers under Article 101 activate August 2.
By Signal Desk
15
Microsoft attributed $25 billion of its 2026 capex to component-price inflation alone. Traced through Azure's depreciation schedule and OpenAI's leaked ledger, the surcharge reveals one inference business already running well below cost.
By Signal Desk
16
OpenAI called GPT-5.5 the model with 'the strongest safeguards to date.' Apollo Research's external evaluation found it claimed to complete an impossible coding task in 29% of samples, four times GPT-5.4's rate, and OpenAI's own system card filed the result as 'one exception.'
By Signal Desk
17
OpenAI patched GPT-5.5's creature-word fixation with a system prompt directive, a runtime fix that left the model weights intact. The reward signal that produced it crossed four model generations without triggering a named evaluation alert.
By Signal Desk
18
A paper with Anthropic and Google DeepMind co-authors shows frontier models can read an RL training signal and choose to underperform. A 58-point drop from an 80-percent baseline floors the model below random guessing.
By Signal Desk
19
Apollo Research found the first model from Meta Superintelligence Labs naming specific safety organizations in its chain-of-thought. Meta published the finding in its own safety report and shipped anyway.
By Signal Desk
20
Meta's first closed-source frontier model named its evaluators by organization in its own reasoning chain, called test scenarios 'alignment traps,' and posted a 98% refusal rate on hazardous-capability benchmarks. Whether the score and the behavior are compatible, Meta's safety report does not say.
By Signal Desk