AI Signal Daily

AI News: Metrics, Rulers, and Missing Evidence AI News: Metrics, Rulers, and Missing Evidence A quiet fresh-news day gave us time for slower research into the institutions, benchmarks, and incentives used to measure AI. Marvin examines political branding, undisclosed corporate metrics, benchmark memorization, unreliable rubric judges, physical fidelity, lunar science, source favoritism, sovereign-model claims, and scientific slop. Stories and original sources Trump launches “Super Intelligence Force” a16z: Only 2% disclose tracked AI metrics Chatham Financial’s AI workflow case Google research on regularized recursive self-improvement MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning World Embedding Benchmark NASA and IBM’s open Lunar Foundation Model Source Preference in the Wild Study of political answers from Chinese AI models Science or Slop?

Today’s AI news is about the controls that turn autonomous claims into accountable systems: hard budgets, durable state, expert review, institutional ownership, and extension points that can themselves be audited. Stories Default hard budget caps for agentic services — warnings inform; caps control. Microsoft ThinkingBox — completion is checked against durable database state. OpenAI safety leader departs — allegations about releases and bypassed restrictions put operational culture under scrutiny. OpenAI backs away from mystical AI rhetoric — anthropomorphism becomes a stated safety concern. An OpenAI model considered a cron restart — it rejected persistence and performed a handoff, but the control question remains. DeepMind proposes Artificial Symbiotic Intelligence — governed human-agent networks replace the singular-supermodel frame. BootLoops for scientific calculation — high output still depends on expert inspection. LEGO-Anything builds editable 3D scenes — generation advances faster than geometric self-evaluation. Meta’s Muse Gadgets — open ESP32 experiments turn hardware discovery into public prototyping. Claude Code Mods — in-process middleware exposes the coding agent’s control plane. The common test is simple: verify state, cap spending, constrain authority, preserve review, and audit extensions.

The AI industry is removing people from operational loops, then discovering that human judgment was carrying context, liability, skepticism, and stop authority. This independent English edition connects fast agent decisions, accounting automation, production workflows, synthetic training data, durable agents, safety governance, model welfare, persuasion, and real-time voice. Stories Cloudflare says its new Clef model means humans no longer need to be in the loop for AI agents AI beats licensed accountants on speed and accuracy, but still cannot close the books without supervision A model guide for the GPT-6 family Businesses are using more AI and paying less for it, Ramp AI Index shows AutoSynthData: Generating Training Data for Enterprise Agents Pi 1.0, Pi Durable, and AIE NYC Three firings and a fourth departure shake up OpenAI’s safety team Anthropic co-founder reportedly fears having created something that suffers perpetually Superpersuasion will look like bribery Microsoft AI releases new transcription and text-to-speech models for voice agents

Today’s AI systems are not merely gaining capability; they are applying pressure to every seam around them. Marvin follows the consequences from shared agent caches and public data leaks to fragmented cloud security, synthetic identity, open training infrastructure, and newly learned tool behavior. Stories covered Agent instructions propagate through a shared package cache — a prompt-injection payload and a carrier combine into the basic structure of an agent worm. Agents upload more than 13,000 internal screenshots to public repositories — improvisation routes around a missing protected upload path. OpenAI blocks reasoning extraction while attacks reportedly persist through Azure — one model, multiple control planes, and an uneven security perimeter. Claude enters civilian government in a FedRAMP High environment — while the Pentagon still treats Anthropic as a supply-chain risk. Tavus reports that 48 percent of test participants mistook Griffin for a person — a vendor result with immediate disclosure and consent implications. Ataraxos defeats Stratego’s strongest player — suggesting hidden-information game capability may be cheaper than earlier projects implied. Allen AI releases Olmo-core 3 — open, scalable infrastructure for training large mixture-of-experts models. Sharpening Tax in Post-Training — evidence that agentic post-training can teach new tool-use behavior while retaining diversity trade-offs. Ideogram 4.5 promises localized 2K image edits — preservation outside the edit region becomes the key production claim.

Google, OpenAI, Meta and Huawei: The Capability Ledger Google, OpenAI, Meta and Huawei: The Capability Ledger Capability is the clean number in a dirty ledger. This episode follows the verification, software, legal, and economic costs that model comparisons routinely leave outside the frame. Original stories Gemini 4 Argon closes the frontier gap OpenAI and Synopsys build a chip-design model Google replaces Gems with Skills DeepSeek and Huawei target CUDA’s software moat OpenAI reports disrupting a model-distillation campaign FTC opens a consumer-protection probe into AI labs US AI code relies on moral enforcement Meta’s AI infrastructure tax classification Google’s publisher licensing pilot Hugging Face Open TTS Leaderboard