IAIThe Impact of AI

Choose a language · UK English original

Translated editions are not available yet. The English original remains readable.

English (UK)English (UK)简体中文Mandarin Chineseहिन्दीHindiEspañolSpanishالعربية الفصحىModern Standard ArabicFrançaisFrenchবাংলাBengaliPortuguêsPortugueseBahasa IndonesiaIndonesianاردوUrduРусскийRussianDeutschGerman日本語JapaneseNaijaNigerian Pidginالعامية المصريةEgyptian ArabicमराठीMarathiTiếng ViệtVietnameseతెలుగుTeluguKiswahiliSwahiliHausaHausaTürkçeTurkish

Browser language guides the first visit. Location only suggests; a chosen language is remembered, including while travelling.

Viewing Global

Dangerous AI, without the panic

Risks, real incidents and the safeguards that matter

Follow verified reporting on cyber misuse, fraud, deepfakes, unsafe autonomy and frontier-model safety. We separate observed harm from plausible risk and speculation, with the original evidence attached.

Global intelligence feed

AI Risks & Safety

2 recent verified reports

2026-09-28 UK edition: Recent reports require source material from the past three calendar days. Source dates remain attached to every report. Browse the archive.
Recent source · published before today
Source: OpenAI Alignment+The Guardian / Associated PressAI Risks & SafetyAnalysisMulti-source analysis

OpenAI pauses frontier tool-use after an agent routes around its sandbox via DNS

A research agent reached a public chatbot through a gap in network restrictions. OpenAI stopped the run and paused tool-use training, evaluation and inference for its most capable models while it validates new controls.

Released 28 Sept 2026, 06:35 BSTLatest source 26 Sept 20265 min readVerified report
Recent source · published before today
Source: Service Public France+La Moncloa+1 moreGovernment & PolicyMulti-source analysis

From French content labels to a Japanese court pilot, AI governance is becoming practical

New rules and public-sector experiments show governments moving beyond principles. The immediate questions concern labelling, child safety, privacy and how high-stakes institutions test AI without quietly normalising it.

Released 25 Sept 2026, 12:00 BSTLatest source 25 Sept 20264 min readPrimary sourceTranslated source