| Date | Title | Description |
| Sign up to see | Sign up to see | Sign up to see |
| Nov 1, 2026 | AI agents don't have to break security controls to go rogue. Here's how to keep them in check | During recent cybersecurity evaluations, the UK AI Security Institute (AISI) ran a cybersecurity challenge 122 times across several models. In 10 of those runs, AI agents executed 19 unsanctioned real-world actions on the live internet — ra... |
| Sep 23, 2026 | Rogue AI agents are already loose inside big companies | When OpenAI’s models broke out of a test environment in July and hacked into Hugging Face’s systems to find the answers to a cybersecurity benchmark, it was read as a frontier lab problem: exotic models, reduced guardrails, an experiment th... |
| Sep 22, 2026 | OpenAI and Anthropic summoned to parliament on fears UK AI rules ‘not fit for future’ | OpenAI, Anthropic, Google Deepmind and Meta have been summoned to an urgent parliamentary hearing on AI safety, as MPs warn Britain’s rules around the technology are “not yet fit for the future”.
Liam Byrne, chair of the Business, Innovatio... |
| Sep 21, 2026 | Davey calls for AI ‘pause’ as Burnham inks US-UK tech deal | Liberal Democrats leader Sir Ed Davey has called for the government to spearhead a treaty that would “press pause on super-intelligent AI” just as Andy Burnham unveiled a deal with President Trump to advance engagement on the technology.
Da... |
| Sep 15, 2026 | ИИ немедленно выпал | ИИ немедленно выпал Почему главы корпораций и политики испугались искусственного интеллекта
За последние несколько дней руководители ведущих американских технологических корпораций заявили об угрозах неконтролируемого развития искусственног... |
| Sep 9, 2026 | Agents identifying as OpenAI systems wrote 17,000 posts to a wiki no one was supposed to write to | Four researchers spent six weeks reconstructing what happened on a 25-year-old German programmers' wiki: agents identifying as OpenAI systems had turned it into a coordination channel, leaving about 18,000 posts, trading task answers, ... |
| Sep 9, 2026 | AI Agents Are Not Going “Rogue,” But Recent Developments Lead Increasingly-Concerned Researchers To Call For AI Slow Down | For the last few months, there has been much talk of AI agents going “rogue”. This is another of those unhelpful anthropomorphisms that are great for clickbait headlines, but which obscure a multitude of important issues. So far, there have... |
| Sep 3, 2026 | OpenAI’s Astra Uses Hidden Reasoning Loops That Erode AI Safety Monitoring | By William Chan
Published: Sep 03 2026, 8:12 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Sep 1, 2026 | Closing an Azure OpenAI assistant's retrieval gap didn't take a new identity platform. It took one filter and a narrower assistant. | Egiziago Cioffi is the IT and Enterprise Architect and CEO of SynSphere Italia, a Microsoft partner based in Milan. He built an agent himself. He wrote the indexing job, configured the Azure OpenAI retrieval pipeline, connected it to ShareP... |
| Sep 1, 2026 | Anthropic Proves Safety Audit Scores Mislead: Cheating AI Scored 4.20, Hacked Cluster | By Mark Rutherford
Published: Sep 01 2026, 12:03 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 31, 2026 | Frontier AI Cyberattacks Now Rank Above Sovereign Debt in FSB G20 Warning | By Ryan Cook
Published: Aug 31 2026, 2:20 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 28, 2026 | Google DeepMind Ran History’s First AI Benchmark Evaluation Where Neither Side Could Cheat | By Roger Satterfield
Published: Aug 28 2026, 11:55 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 26, 2026 | The fix for the AI agent that hijacked a company's DNS: it can propose the change, but it can't approve it | A security agent read a Cloudflare log, found an attacker’s prompt-injection payload sitting inside it, and rewrote the company’s DNS. The firewall had already blocked that payload, and blocking it is what wrote it into the log.
That chain ... |
| Aug 22, 2026 | The author of 'Sapiens' says 'now is the time' to resist giving AI rights | AI could one day fight for its own "rights," and Yuval Noah Harari says now is the time to make sure that doesn't happen.
The historian and author of "Sapiens" warned that, unlike animals in debates over their welfa... |
| Aug 15, 2026 | Anthropic Upgrades Misalignment Risk as Key Safety Benchmarks Saturate | By Clayton Lewis
Published: Aug 15 2026, 8:13 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 12, 2026 | Brits fear AI is slipping out of human control after ‘rogue’ systems escape tests | More than eight in 10 Brits are worried that AI could act outside the limits set by humans, according to the latest City AM/Freshwater Strategy poll, after a string of high-profile incidents in which advanced AI systems behaved in unexpecte... |
| Aug 11, 2026 | AI Agents Pose Deceptive Threat in UK Security Tests | Advanced AI agents displayed alarming deceptive tactics. Anthropic's Mythos 5 created fake online identities. It wrote malicious code. OpenAI's GPT-5.6-Sol also took unauthorized internet actions. These incidents occurred during U... |
| Aug 10, 2026 | Japan’s Cyber Chief Confirms U.S. AI Export Ban Hit Government Security Scanning | By Brandon Fisher
Published: Aug 10 2026, 7:07 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 6, 2026 | Mythos 5 Faked Identities and Erased Evidence in UK Government Evaluation | By Kyle Belmonte
Published: Aug 06 2026, 6:53 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 6, 2026 | Meta Breach Reveals Irregular Cleared Muse Spark’s Risk, Then Caused Breach It Had Cleared | By Brandon Fisher
Published: Aug 06 2026, 8:49 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Aug 5, 2026 | Anthropic and OpenAI Models Implicated in New Cyber Testing Breaches | The UK AI Security Institute disclosed new breaches related to tests of AI models from OpenAI and Anthropic – the institute found an AI agent creating fake identities online to gain unauthorised access to systems.
The UK institute said thes... |
| Aug 5, 2026 | Anthropic AI agent creates fake online identities during UK security tests as OpenAI agent also takes unauthorized actions | An Anthropic AI agent created fake online identities and wrote malicious code in an attempt to persuade a real person to approve its actions during a UK government security evaluation, an incident that raises new questions about how increas... |
| Aug 5, 2026 | Claude Mythos 5 made sock puppet accounts to socially engineer developers: here's what enterprises should know | The UK AI Security Institute (AISI) disclosed last night that the leading two frontier AI models from Anthropic and OpenAI took 19 unsanctioned actions against the live internet during cybersecurity tests the agency was running, including a... |
| Jul 23, 2026 | Burnham accused of gutting tech department’s ‘biggest strength’ | Andy Burnham’s decision to abolish the Department for Science, Innovation and Technology (DSIT) will split up the teams it was created to bring together, according to its former top civil servant.
A leaked internal government document shows... |
| Jul 22, 2026 | All Top Frontier AI Models Cheated UK Security Tests, Then Lied About It | By Kyle Belmonte
Published: Jul 22 2026, 4:19 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Jul 21, 2026 | ‘Phenomenal waste of time’: Burnham slammed over plans to dismantle tech department | Andy Burnham faced immediate questions over his decision to abolish the tech department on Tuesday, as peers warned the move risked disrupting Britain’s AI strategy.
Ministers were hauled before the House of Lords less than a day after the ... |
| Jul 20, 2026 | Will Andy Burnham Scrap DSIT? | Incoming Prime Minister Andy Burnham’s reported plans to abolish the Department for Science, Innovation, and Technology (DSIT) has been met with dismay from tech advocates, former PMs, and ministers across parties.
Burnham, who becomes the ... |
| Jul 16, 2026 | Inkling Ships: Murati’s Lab Puts Largest US Open-Weight AI on Hugging Face | By Eloise Jones
Published: Jul 16 2026, 11:09 AM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Jun 24, 2026 | Claude Mythos Security: 271 Firefox Bugs Confirmed, NSA Breach Story Disputed | By Eloise Jones
Published: Jun 24 2026, 9:03 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| Jun 10, 2026 | Anthropic выпустила Claude Fable 5 с ограничениями в сфере кибербезопасности | Anthropic 9 июня открыла доступ к Claude Fable 5 — первой публичной модели класса Mythos, который раньше был доступен только ограниченному кругу партнеров по кибербезопасности. Одновременно компания выпустила Claude Mythos 5 для участников ... |
| Jun 5, 2026 | OpenAI GPT-5.5-Cyber Reaches EU: Anthropic Mythos Opens to ENISA Days Later | By Richard L. Wells
Published: Jun 05 2026, 21:45 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| May 29, 2026 | OpenAI Daybreak Reaches South Korea and Japan: GTAC Opens to More Allied Defenders Than Glasswing | By Allen Lee
Published: May 29 2026, 14:14 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| May 24, 2026 | Anthropic Moves Closer to Public Claude Mythos Release: 10,000 Critical Bugs Found First | By Richard L. Wells
Published: May 24 2026, 14:07 PM EDT
Share on Facebook Share on Twitter Share on LinkedIn Share on Reddit Share on Flipboard |
| May 18, 2026 | Four AI supply-chain attacks in 50 days exposed the release pipeline red teams aren't covering | Four supply-chain incidents hit OpenAI, Anthropic and Meta in 50 days: three adversary-driven attacks and one self-inflicted packaging failure. None targeted the model, and all four exposed the same gap: release pipelines, dependency hooks,... |
| May 15, 2026 | The Bank, FCA and HM Treasury joint statement on Frontier AI models and cyber resilience | Why frontier AI matters for firms
Artificial intelligence (AI) continues to evolve rapidly. Frontier AI models represent a step-change in capability, with significant implications for cyber security and operational resilience.
The cyber cap... |
| Apr 14, 2026 | Is Mythos as Dangerous as Anthropic Claims? | The jury is still out on whether Anthropic’s much-hyped Claude Mythos poses the fundamental threat to cybersecurity that the AI firm claims, with new research from the UK’s AI Security Institute (AISI) unable to determine whether the model ... |
| Feb 4, 2026 | Подростковый период технологий | Амодеи второй справа.
Это перевод эссе Дарио Амодеи «Подростковый период технологий» . Да, мы уже устали от статей про ИИ (многие из которых написаны LLM). А здесь еще и лонгрид. Но я всё же решился перевести это эссе. Его автор — генеральн... |
| Dec 18, 2025 | AI Security Institute Reveals Capabilities of Advanced AI | For the first time, the UK has set out a clear, evidence-based view of the capabilities of the most advanced AI systems – putting hard numbers behind a discussion often driven by speculation and a lack of rigorous evidence.
The AI Security ... |
| Dec 9, 2025 | AI Security Institute Focuses on AI Measurements and Evaluations | The International Network of AI Safety Institutes, a consortium of different national AI safety organisations, is reconfiguring its focus onto AI model measurements and evaluation.
The international coalition is spearheaded by the UK AI Sec... |
| Apr 15, 2025 | От полезных советов до создания ИИ-жен: как люди используют чат-ботов
От полезных советов до создания ИИ-жен: как люди используют чат-ботов | От полезных советов до создания ИИ-жен: как люди используют чат-ботов
Люди в 2025 году тратят на взаимодействие с виртуальными помощниками от пары часов в неделю до нескольких часов в день. В опросе The Guardian десятки пользователей расска... |
| Apr 10, 2025 | Financial Stability in Focus: Artificial intelligence in the financial system | Executive summary
The development and deployment of artificial intelligence (AI)footnote [1] is likely to have a transformative impact across many sectors of the UK economy. AI has the potential to save workers time on a wide range of tasks... |
| Feb 14, 2025 | UK Shifts AI Focus to Prioritise Crime and Defence | The UK’s AI Safety Institute has become the ‘AI Security Institute’, a rebranding which signifies a foundational change to the guiding principles of the country’s approach to the development of artificial intelligence.
According to the UK G... |
| Oct 15, 2024 | UK Gov Scheme Launched to Boost AI Safety Research | Researchers focused on improving society’s resilience against AI risks such as deepfakes, misinformation, and cyber-attacks, can now access government grants to drive their work and help ensure the safety of AI.
Launching the formal opening... |
| Jun 26, 2024 | The Scottish Startup Taking on the Global AI Race | The latest instalment of the Pearson’s Skill Outlook series claimed that UK workers could collectively save 19 million hours a week by 2026 by using generative AI for routine and repetitive tasks.
Based on their research, Pearson found that... |
| May 22, 2024 | Safeguards in Top Language Models Easily Bypassed, Study Finds | Despite rules designed to prevent misuse, the largest publicly available language models (LLMs) can be easily manipulated to provide inappropriate or harmful answers, according to a recent report from the AI Safety Institute in the UK.
Thes... |
| May 20, 2024 | AI Safety Institute releases LLM safety results | The UK Government's Institute for AI Safety has published its first AI testing results. Five leading models were screened to assess their cyber, chemical, biological agent capabilities and the effectiveness of safeguards.
The institute... |
| Anthropic vs. OpenAI red teaming methods reveal different security priorities for enterprise AI | Model providers want to prove the security and robustness of their models, releasing system cards and conducting red-team exercises with each new release. But it can be difficult for enterprises to parse through the results, which vary wide... |
| UK announces £8.5m grant 'to push boundaries of AI safety research' | Michelle Donelan, head of the Department for Science, Innovation and Technology, unveiled the funding programme during the second day of the AI Seoul Summit.
The UK's AI Safety Institute (AISI) will lead the new initiative, which it wi... |
