Kohli steps downs as India Test captain

• AI cyber capabilities are now doubling every four months, government-backed tests show.
• New models can find vulnerabilities and generate exploit code without human expertise.
• Businesses of all sizes, not just critical sectors, are now potential targets.

The UK government has warned that artificial intelligence is rapidly lowering the barrier to cyberattacks, allowing systems to identify software weaknesses and generate exploit code at a speed and scale not seen before. In a joint letter issued on April 15, 2026, ministers said the shift marks a move away from attacks led by a small pool of highly skilled criminals to a landscape where advanced tools can replicate those capabilities. The warning follows testing of Anthropic’s Mythos model, which officials said is “substantially more capable” at cyber offence than earlier systems, as quoted in the letter.

Data from the government-backed AI Security Institute suggests the pace of change is accelerating sharply, with frontier AI cyber capabilities now doubling roughly every four months, compared to every eight months earlier. Ministers also pointed to parallel developments across the industry as evidence that the shift is not isolated, adding that attackers are expected to target “ordinary companies, of every size, in every sector." Against this backdrop, the government has set out three immediate steps for businesses to strengthen their defences.