AI as an accelerator of attacks, Iran offline for several days and million-pound fines in the UK
The digital space in recent days shows three different forms of power: AI as an "accelerator" of an attacker's work, the ability of a state to restrict connectivity at the level of an entire country, and strict enforcement of online child protection.
The digital space in recent days shows three different forms of power: AI as an "accelerator" of an attacker's work, the ability of a state to restrict connectivity at the level of an entire country, and strict enforcement of online child protection.
AI and attacks on Mexican public institutions: what researchers claim (and what the authorities say)
Security researchers describe a case in which an unknown attacker allegedly used the Claude model (Anthropic) as an assistant to find weaknesses, write scripts, and automate steps in attacks on Mexican public institutions.
According to these findings, roughly 150 GB of data were allegedly stolen, including documents related to tax and electoral records and other sensitive files. But this is important: Mexican institutions, according to available statements, deny the breach or say they did not find evidence of an intrusion in their logs. In other words — this is a situation where the public story is composed of researchers' claims, media findings, and conflicting responses from some of the affected authorities.
Source (summary of research + responses from authorities): https://www.latimes.com/business/story/2026-02-26/hacker-used-anthropics-claude-ai-to-steal-mexican-government-data
Iran offline: the blackout has lasted several days
This is not a short outage. Major media already described a "near-total" blackout on 28 February 2026 with connectivity around 4% of normal levels (citing NetBlocks monitoring). There are also follow-up reports indicating that the restrictions continued in the following days — including reports that the blackout entered its fourth day on Tuesday 3 March.
What is important for the technical context:
- such a deep drop in connectivity usually corresponds to interventions at the transit / backbone routing level and coordinated restrictions across operators, not just "blocking apps";
- a prolonged blackout significantly complicates verification of information, internal communication within the country, and the work of the media.
Sources:
- 2026 (blackout as a tool of control, ongoing impacts): https://www.theguardian.com/world/2026/mar/02/internet-blackout-regime-iranians-experts-digital-censorship
- 2026 (blackout "entered fourth day", citing NetBlocks): https://www.iranintl.com/en/202603031305
Britain: million-pound fines over age verification and children's data
British regulators have switched from warnings to enforcement.
Ofcom fined company 8579 LLC £1.35 million for failing to implement "highly effective" age verification on services with pornographic content. The decision also includes an additional fine for ignoring an information notice and the threat of ongoing daily penalties if the company does not remediate. https://www.ofcom.org.uk/online-safety/protecting-children/ofcom-fines-porn-company-1.35-million-for-not-having-age-checks
At the same time, the British ICO fined Reddit £14.47 million for failures to protect children's data and shortcomings in age assurance, including criticism that the company did not carry out the required Data Protection Impact Assessment (DPIA) in time. Reddit has announced it will appeal. https://cy.ico.org.uk/about-the-ico/media-centre/news-and-blogs/2026/02/reddit-issued-with-1447m-fine-for-children-s-privacy-failures/