CDT Europe’s April 2026 AI Bulletin, published 28 April, reports that the final political trilogue on the AI Omnibus package took place on 28 April 2026, with the last open question being whether high-risk AI systems embedded in products regulated …
Singapore’s Cyber Security Agency published Advisory AD-2026-004 on frontier AI cybersecurity risks, noting that frontier models can reduce the time required to identify vulnerabilities and engineer exploits from months to hours. The advisory sets out hardening measures for organisations, including …
The People’s Armed Police Force has shown how drones, uncrewed armoured vehicles and robot dogs could autonomously contain urban unrest.A new study from mainland China’s internal security forces offers a glimpse into a future where urban unrest is managed by …
The San Diego Police Department has formally barred officers from using any generative AI tool to author police reports unless specifically authorised by the department. The ban reflects California Senate Bill 524 (effective January 2026), which requires agencies to disclose …
At TED2026 in Vancouver on 13 April, Flock Safety CEO Garrett Langley argued that AI-powered ALPR and drone technology replaces subjective judgement with evidence-based policing, citing more than a million crimes solved annually through Flock’s network. Langley addressed constitutional and …
The Metropolitan Police used Palantir’s data analytics platform to cross-reference internal records including sickness, overtime, expenses and building access. The exercise identified hundreds of officers facing gross misconduct proceedings for corruption, fraud and abuse of authority.
A Washington Post investigation examines Longeye and similar AI platforms adopted by US police departments to process case documents faster. Courts have not settled on disclosure obligations when AI analysis forms part of the investigative chain.
A 23 April 2026 analysis warns that organisations pausing AI Act compliance in anticipation of the reported Omnibus extension risk being caught without documentation or conformity assessment readiness if the extension does not clear formal OJ publication before 2 August …
Microsoft published a Security Blog post on 22 April 2026 directly responding to the capabilities demonstrated by Anthropic’s Claude Mythos Preview. The post acknowledges that AI models can now autonomously discover weaknesses, chain multiple lower-severity issues into working exploits and …
Stanford HAI’s 2026 AI Index Report, published this week, documents a widening gap between frontier AI capability benchmarks and responsible AI safety benchmarks. Almost every frontier model developer reports results on capability benchmarks; the same is not true for responsible …
CDT Europe’s April 2026 AI Bulletin, published 28 April, reports that the final political trilogue on the AI Omnibus package took place on 28 April 2026, with the last open question being whether high-risk AI systems embedded in products regulated …
Singapore’s Cyber Security Agency published Advisory AD-2026-004 on frontier AI cybersecurity risks, noting that frontier models can reduce the time required to identify vulnerabilities and engineer exploits from months to hours. The advisory sets out hardening measures for organisations, including …
The People’s Armed Police Force has shown how drones, uncrewed armoured vehicles and robot dogs could autonomously contain urban unrest.A new study from mainland China’s internal security forces offers a glimpse into a future where urban unrest is managed by …
The San Diego Police Department has formally barred officers from using any generative AI tool to author police reports unless specifically authorised by the department. The ban reflects California Senate Bill 524 (effective January 2026), which requires agencies to disclose …
At TED2026 in Vancouver on 13 April, Flock Safety CEO Garrett Langley argued that AI-powered ALPR and drone technology replaces subjective judgement with evidence-based policing, citing more than a million crimes solved annually through Flock’s network. Langley addressed constitutional and …
The Metropolitan Police used Palantir’s data analytics platform to cross-reference internal records including sickness, overtime, expenses and building access. The exercise identified hundreds of officers facing gross misconduct proceedings for corruption, fraud and abuse of authority. Scotland Yard is now …
A Washington Post investigation examines Longeye and similar AI platforms adopted by US police departments to process case documents faster. Courts have not settled on disclosure obligations when AI analysis forms part of the investigative chain. Defence attorneys argue they …
A 23 April 2026 analysis warns that organisations pausing AI Act compliance in anticipation of the reported Omnibus extension risk being caught without documentation or conformity assessment readiness if the extension does not clear formal OJ publication before 2 August …
Microsoft published a Security Blog post on 22 April 2026 directly responding to the capabilities demonstrated by Anthropic’s Claude Mythos Preview. The post acknowledges that AI models can now autonomously discover weaknesses, chain multiple lower-severity issues into working exploits and …
Stanford HAI’s 2026 AI Index Report, published this week, documents a widening gap between frontier AI capability benchmarks and responsible AI safety benchmarks. Almost every frontier model developer reports results on capability benchmarks; the same is not true for responsible …