
📡 Your radar for the AI era. Daily news on AI, cybersecurity, privacy, SIAM and agile, plus the certifications that grow your career. #EXINRadar
Utrecht, the Netherlands
Joined July 2009
- Tweets3.3K
- Following3.5K
- Followers3.1K
- Likes614
UK regulator Ofcom left device scanning out of new child safety rules.
Large platforms must plan to detect previously unseen child abuse material in public posts from February 2027 if Parliament approves.
therecord.media/uk-ofcom-chi…
6
Apple added more granular app permissions in macOS 27.3 beta 2. Existing grants remain valid.
Developers should support access limited to specific folders or apps for new permission requests.
helpnetsecurity.com/2026/11/…
30
FAA officials traced Friday's nationwide ground stop to an Arkansas router. Disruption hit 17 of 22 en route centers.
Network teams should review cross-site disruption risks during network upgrades.
thenextweb.com/news/faa-grou…
14
The US Commerce Department issued rules requiring licenses for advanced chip exports to Indonesia, Malaysia, Singapore, Thailand and Brazil.
US-headquartered firms must file monthly reports to qualify for an exemption.
thenextweb.com/news/us-remot…
15
Netskope researchers found regulated data accounted for 57% of AI-related data policy violations in manufacturing.
DLP controls should check sensitive data before it reaches personal app instances or AI apps.
industrialcyber.co/manufactu…
10
Google’s Q4 AI Threat Tracker found gateways called LLM Mixers selling stolen AI access; business token prices ranged from $0.10 to $25.
Set token limits and usage-spike alerts across AI accounts to limit compromise costs.
zdnet.com/innovation/google-…
1
1
29
OpenAI will embed six Apollo Research safety evaluators in a pilot to test whether AI models undermine safeguards.
Risk teams should note OpenAI can limit public evidence supporting the pilot’s independent conclusions.
techcrunch.com/2026/11/16/op…
11
The UAE signed the San Francisco Declaration to safeguard sensitive AI technology and critical infrastructure.
Risk teams should assess theft and illicit-transfer risks for AI software, knowledge and inputs.
thenationalnews.com/future/t…
20
Xint.io researchers report AI model GPT 6 Sol found a Google V8 flaw in September for $4.24 in 13 minutes in selected files.
Security teams can run similar checks, but models may need guidance and extra runs, increasing costs.
zdnet.com/innovation/ai-find…
18
PA Consulting's study found 72% of senior leaders report higher-quality work with AI, versus 43% of entry-level or junior staff.
itpro.com/business/business-…
14
Microsoft traced its 12 November Microsoft 365 outage to a gateway defect, increased load and reduced capacity.
Customers can check the Microsoft 365 Admin Center for the full root cause.
itpro.com/software/microsoft…
1
47
The Department of Energy ran Liberty Eclipse, a utility cyber exercise, on a live, isolated grid on Plum Island, New York.
Utilities can address communication gaps exposed by the exercise before a crisis.
industrialcyber.co/news/does…
1
25
OpenAI's AI model GPT-6 Astra helped produce a proof that could solve a 50-year-old particle physics problem. Physicist Nathaniel Craig says much remains before realistic simulations are possible.
the-decoder.com/openais-gpt-…
15
Attackers used GeoVision camera management software to break into Windows machines at 13 organizations and install cryptominers, Huntress warns. Check for command shells or PowerShell sessions started by the GeoVision service.
helpnetsecurity.com/2026/11/…
16
Tobacco group Imperial Brands plans 2,100 job cuts this year and 2,100 to 2,500 more by the end of 2030 as it uses more AI.
Workforce plans need to account for its Langenhagen factory closure next year.
theguardian.com/business/202…
31
A scan of 382 UK councils by Ronald Petrlic, IT security professor at Nuremberg Institute of Technology, found almost all publish a DMARC policy to stop email forgery, and 60% reject forged mail outright.
computerweekly.com/news/3666…
26
Microsoft has released emergency out-of-band Windows updates to fix activation and licensing failures caused by this month’s security updates, affecting organizations that use the SLMGR tool to automate activation.
bleepingcomputer.com/news/mi…
27
Nokia, Nscale and Swedish telco Telia are among the founding members of a new Nordic consortium to develop “AI Gigafactories”, formed in response to a call from EU leaders.
itpro.com/infrastructure/nok…
43
Reuters found China aims to make every schoolchild AI-literate by 2030, but lacks a standard way to measure it.
thenextweb.com/news/china-ai…
20
Anthropic says Claude Fable 5.2 makes about 60% fewer unnecessary blocks on cybersecurity requests than Fable 5.1, while Claude Mythos 5.2, available to selected partners, is better at finding vulnerabilities.
bleepingcomputer.com/news/ar…
27