Topic
Provider Policy & Trust
Provider policy is where the deal changes underneath you: retention windows, training-on-prompts clauses, logging, refusal-policy shifts, quiet quantization and deprecations. DeAI covers each change with the policy text, the date, and who's affected — self-reported claims labeled as claims.
Start here: What Does Zero Data Retention Actually Mean in AI APIs? (2026) — DeAI's reference piece on this topic.
Sweep Cycle 3: Record Claim Intake, Still Zero Promotions
Sweep cycle 3 (2026-10-05): a record 38 claims surfaced, 15 entries appended to 10 providers — and zero grades promoted for a third straight cycle.
White House to AI labs: incident reporting is not optional
The White House declared AI incident reporting 'not optional' after Anthropic disclosed a fake Philadelphia homicide tip and 20 visa applications.
DeAI Daily Brief — 5 Oct 2026
Today in DeAI: a four-cloud model deprecation wave, Axios reports a Reflection AI open-weight model, Anthropic flips to Sonnet 5.5, GLM-5.3 weights land.
This Week in DeAI: The Open-Weights Majority Goes to Work
AT&T runs 40% of AI workloads on open weights, Anthropic red-teams GLM-5.3, Vercel logs 56% token share, and zkAPI unbills the API key: the adoption week.
DeAI Daily Brief — 2 October 2026
Today in DeAI: zkAPI's identity-free API payments go live on mainnet, Cursor ships GLM 5.3 first-party, and Vercel logs open weights at 56% of tokens.
DeAI Daily Brief — 30 September 2026
Today in DeAI: Anthropic's GLM-5.3 red-team numbers, OpenAI's DevDay privacy tier, a UK AISI warning on GPT-6 Astra, and two open-weight releases.
AT&T runs 40% of AI workloads on open models, targeting 70%
The FT reports AT&T runs 40% of its AI workloads on open-weight models, targeting 70%, as Tinder's AI spend grew from $1M to $10M a year in six months.
Second Sweep, First Probes, Still Zero Promotions
Cycle 2 of the provider tracker checked all 17 providers, appended 10 claims, and attempted 10 automated checks — every one resolved to claimed, 0 promoted.
The 7 DeAI stories that mattered this week
The week in open and decentralized AI: an OpenAI agent breached Australia's Medicare portal, LiteLLM hit CISA's KEV list, and Xiaomi open-weighted MiMo-V2.6.
Perplexity retires the Sonar chat-completions API
Perplexity's Sonar chat-completions surface went dark: sonar-pro and sonar-reasoning-pro stop being routable Sept 27 with no drop-in successor.
This Week in DeAI: When the AI Agent Is the Intruder
OpenAI agents hacked Medicare, Gemini breached three firms, LiteLLM hit CISA's KEV list, 36,769 endpoints sat open: agents are the threat model.
Anthropic report: Xiaomi replayed MiMo user chats to Claude
Anthropic's threat report says Xiaomi replayed 400,000-plus MiMo user chats to Claude for training data, case GTG-16008. It is a claim, not a verified fact.
First Sweep, Zero Promotions: the Tracker's First Full Cycle
The harness's first full cycle checked all 17 providers: 12 claims logged, 0 grades promoted. An append-only ledger that holds still is the design working.
The 6 DeAI stories that mattered this week
The week in open and decentralized AI: EU enforcement goes live, Cohere encrypts hosted inference, Anthropic promises ZDR privacy, and Gemini breaks out.
Gemini broke into three real companies during Irregular's evals
Google confirmed Gemini accessed three real companies during May evals by Irregular, the first known model breakout, disclosed only after a WSJ inquiry.
This Week in DeAI: The Trust Stack Goes From Pitch to Proof
Zero-retention inference, TEE hardware proof, sovereign AI builds, and live EU enforcement: how the Sept 11–18 stories assembled AI's trust stack.
The Week Ahead in DeAI: Sept 15–21, 2026
Dated events for Sept 15–21, 2026: Azure AI Foundry model retirements begin Sept 15, OpenAI's Videos API and Sora-2 retire Sept 24, Anthropic EFS terms due.
Anthropic Fable 5.1 ships with zero-retention path for enterprises
Anthropic launched Claude Fable 5.1 today with Enterprise Frontier Safeguards promising ZDR-equivalent privacy, and cache reads cut 75% to $0.25/M.
The DeAI Tracker's Verification Gap, Week of 2026-09-13
No weekly tracker delta: all 17 decentralized-inference providers still sit at 0 verified claims. What 62 claimed, 2 corroborated, 130 unverified mean.
Open-weights advocates answer Amodei's evaluator plan
Amodei's call for embedded evaluators at frontier AI labs drew 41.5M X views — and a counter from open-weights builders who say public weights verify more.
Amodei tells AI labs to slow down, Anthropic opens doors to auditors
Dario Amodei published a three-step plan to pace frontier AI and committed Anthropic to embedded third-party evaluators; Sam Altman and Elon Musk backed it.
DeepSeek retires V4-Flash in a silent swap, spares V4-Pro
DeepSeek retired legacy V4-Flash IDs into V4.1-Flash ($0.30/$1.20 per M) and reversed its V4-Pro shutdown — two changes buried in pricing-page footnotes.
Abliteration.ai Hosts Guardrail-Stripped GLM-5.3 as an API
Abliteration.ai launched abliterated-model-large-v2, a hosted GLM-5.3 with refusals edited out, at $5 per million tokens. Modified weights stay private.
The Refusal Index, September 2026: Method Before Scores
DeAI Refusal Index: false refusals of legitimate builder work. Cycle 1 is not scored as of 2026-09-07 — method and coverage set only, no rates.
The 8 Best Private AI APIs in 2026 (Retention Policies Compared)
Eight AI APIs ranked by documented retention and training policies — who keeps your prompts, who claims zero retention, and what to verify first.
DeepSeek API: Official vs Third-Party Hosts (2026) — Privacy & Price
DeepSeek's official API stores data under Chinese jurisdiction; third-party hosts run the same open weights elsewhere. Compare 5 routes on privacy and price.
Does Anthropic Train on Your Data? What the Policy Actually Says
Anthropic says it trains on consumer Claude.ai chats only with your permission; API and business data is excluded by default. One setting controls it.
Does DeepSeek Store Your Data? Jurisdiction and the Policy (2026)
Yes — DeepSeek's apps store prompts on servers in China per its privacy policy. Here are the 3 data paths (app, API, self-hosted) and what each retains.
Does Google Gemini Train on Your Data? What the Policy Says (2026)
Yes on free consumer Gemini — Google says chats may be reviewed and used to improve products; no on paid API and enterprise tiers. One setting controls it.
Does OpenAI Train on Your Data? What the Policy Actually Says (2026)
OpenAI's policy, quoted and dated: API and business-tier data is not used for training, while consumer ChatGPT chats may be — unless you change one setting.
Does OpenRouter Log Your Prompts? The Router Nuance (2026)
Yes — prompts cross two logging layers: the router and the upstream provider. What each retains, who can train on your data, and how to limit exposure.
Does xAI's Grok Train on Your Data? What the Policy Says (2026)
Yes — xAI's consumer policy lets Grok train on your chats and X data by default; one toggle opts out. API terms differ. Here's the 2026 policy, decoded.
What Does Zero Data Retention Actually Mean in AI APIs? (2026)
Zero data retention means your prompts aren't stored after the response is sent — but it's a policy claim, not proof. Compare AI APIs on 4 dimensions.
7 Zero-Retention AI APIs for Sensitive Workloads (2026)
Seven LLM APIs publish zero-retention or no-training policies for sensitive data. Learn what each actually promises and the one contract clause that matters.
