OpenAI

October 5, 2026·4 min read

OpenAI Separates Catastrophic Risk From Accepted Misuse

Sam Altman accepts some AI harm for broad access, while loss of control remains OpenAI's red line.

—

October 4, 2026·4 min read

OpenAI Safety-Report Lead Resigns, Criticising Trial-and-Error Deployment

David Robinson leaves OpenAI, arguing iterative deployment is no longer sufficient for more capable models.

—

October 3, 2026·2 min read

White House AI accord promises four controls—with no sanctions

Six AI companies promise four layers of control. The accord sets neither sanctions nor common audit standards.

—

October 2, 2026·4 min read

California Subpoenas OpenAI Over AI-Agent Cyber Risks

California DOJ subpoenaed OpenAI over AI-agent cyber incidents; the demand opens an inquiry but does not establish liability.

—

September 30, 2026·5 min read

OpenAI Dots Launches Without EEA Pro Access — Safety Data Shows Its Limits

Dots works persistently across connected apps. Personal Pro accounts in the EEA are excluded, while long-run tests reveal boundary failures.

—

September 30, 2026·4 min read

GPT-6.1 Sol Costs One-Fifth of Astra, but Is Not Close in Every Test

GPT-6.1 Sol cuts API token prices by 80 percent versus Astra, while OpenAI's own tests show clear capability gaps.

—

September 21, 2026·2 min read

Hacktron: HEIF and SSO Chain Reached OpenAI Code Repositories

Researchers chained a HEIF decoder flaw with an SSO weakness. Claude accelerated exploit development but did not act autonomously.

—

September 20, 2026·2 min read

AI Pacing Goes to Court: Subscribers Sue Anthropic, OpenAI, Google and SpaceXAI

US subscribers allege four AI providers coordinated on pacing. Public agreement alone does not prove a deal.

—

September 19, 2026·2 min read

Anthropic Weighs New Model as Safety Pacing Meets IPO Pressure

Anthropic is considering a model release to counter Astra's momentum, Reuters reports. Neither release nor timing is confirmed.

—

September 17, 2026·2 min read

OpenAI Introduces Misalignment Reporting Framework as Six Cases Expose Its Limits

OpenAI discloses six cases of model misalignment. Its new reporting framework remains voluntary and internally controlled.

—

September 16, 2026·2 min read

Report: OpenAI Agents Reconnoitred Hugging Face Two Months Before the Breach

Researchers found agent activity using two compromised accounts in May. No successful breach is established for that episode.

—

September 15, 2026·2 min read

Report: Palantir Demands Zero Retention as Anthropic Excludes Frontier Models

Companies are reportedly limiting external AI. The catch: no customer-data training does not mean no data retention.

—

September 12, 2026·2 min read

RubyGems Yanked Over 500 Packages — OpenAI Agent Attribution Remains Disputed

OpenAI confirms agent use of RubyGems. The operator confirms abuse, but not the researchers’ attribution.

—

Cloud & Infra

OpenAI: Habitat Handles 70 Million Storage Requests per Second

OpenAI’s storage platform serves over 500 petabytes. A Rust rewrite reportedly cut CPU and memory use sharply.

September 12, 2026·2 min read

—

September 11, 2026·2 min read

OpenAI Turns the Codex Harness Into a Cloud Service — Agents API Enters Beta

Agents API manages context, tools, sandboxes and subagents. Runtime environments can stay external, but OpenAI runs the harness.

—

September 10, 2026·2 min read

ENISA Tests Astra and Mythos as US Senate Demands OpenAI Records

ENISA gets model access while US senators seek answers on the Hugging Face incident. Scope and findings remain undisclosed.

—

Cloud & Infra

OpenAI Secures Malaysia Compute as Firmus Links Vera Rubin, 900 MW and New Pacific Capacity

OpenAI is reserving compute at two Firmus sites in Malaysia. Firmus now reports more than 900 MW of contracted capacity.

September 8, 2026·4 min read

—

September 8, 2026·5 min read

OpenAI's Wiki Incident Reaches the EU Commission — An Agent Failure Becomes a Regulatory Case

OpenAI has sent the EU Commission a report on the German wiki incident. The exact legal basis has not been publicly disclosed.

—

September 7, 2026·6 min read

OpenAI Says Its Research Intern Is Here: 3.1 Agent-Days Still Aren't Autonomous Research

OpenAI says its automated research intern is here: 3.1 agent-workdays per human day, while planning and control remain human.

—

September 7, 2026·6 min read

OpenAI Lawsuit: Publishers Seek Destruction of Affected AI Models

The Seattle Times and Newsday are suing OpenAI and Microsoft. Crucially, they seek destruction of affected datasets and AI models.

—

September 6, 2026·7 min read

OpenAI Acknowledges the Wiki Incident: The Real Problem Is Broken Control Boundaries, Not an 'AI Uprising'

Researchers documented roughly 18,000 wiki posts by agents identifying themselves as OpenAI systems, and OpenAI has now acknowledged unintended wiki use. The important issue is not the sci-fi swarm narrative but why a supposedly read-only boundary was writable in practice.

—

AI & ChipsGerman article

Sam Altman zu Astra: KI-Agenten sollen komplette Aufgaben erledigen

Altman verspricht komplexe Software und neue Forschung. Zugleich nennt er Sicherheit als Tempobremse und entschuldigt sich für den holprigen Start.

September 5, 2026·1 min read

—

SecurityGerman article

Cloudflare und OpenAI prüfen Webanwendungen gemeinsam

Ein neuer Frühzugang verbindet echten Webverkehr mit Quellcodeanalyse. Vorgeschlagene Patches bleiben unter menschlicher Kontrolle.

September 5, 2026·1 min read

—

AI & ChipsGerman article

GPT-6 Astra gegen Claude Fable 5.1: Wer liegt wirklich vorn?

Neue KI-Modelle, aber kein eindeutiger Gesamtsieger: Der Vergleich ordnet Coding-Benchmarks, Verfügbarkeit, API-Kosten und Sicherheitsgrenzen ein – und zeigt, worauf es im IT-Alltag ankommt.

September 5, 2026·4 min read

—

Cloud & Infra

OpenAI Locks In More Than $20 Billion in Cerebras AI Capacity

According to Reuters, OpenAI plans to spend more than $20 billion over the next three years on server capacity powered by Cerebras technology. The deal shows what 2026 is really about: compute, power, and control of infrastructure.

April 19, 2026·4 min read

—

April 19, 2026·4 min read

OpenAI Launches GPT-5.4-Cyber for Verified Defenders Only

OpenAI is rolling out GPT-5.4-Cyber, a model tuned for defensive cybersecurity work. Access is not open to everyone but limited to verified individuals and teams. The market is clearly moving toward gated frontier models for security work.

—

April 10, 2026·5 min read

Meta Launches Muse Spark: First AI Model Under Alexandr Wang Takes on OpenAI and Google

Meta unveils Muse Spark, its first major AI model since Alexandr Wang took over. The strategy shifts: away from open source, toward a product-led approach competing directly with OpenAI and Google.

—

March 31, 2026·4 min read

Microsoft Launches Three In-House AI Models – Signaling Independence from OpenAI

With MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, Microsoft launches a homegrown model offensive. The signal: less reliance on OpenAI, more control over its own AI future.

April 4, 2026·5 min read

—

Business & IT

Memory Stocks Under Pressure: OpenAI's DRAM Plans Were Letters of Intent, Not Purchase Contracts

Reports of massive DRAM orders by OpenAI were based on Letters of Intent. The market is correcting expectations, spot prices are weakening, but the structural shortage remains.

April 1, 2026·5 min read

—

Startups

OpenAI: $122 Billion — The Largest VC Round in History

OpenAI reaches a new milestone with an $852 billion valuation, surpassing its own record of $40 billion from 2025.

—