Russian-state hackers exploited a critical Microsoft Office vulnerability (CVE-2026-21509) less than 48 hours after Microsoft released an urgent security update late last month, according to researchers at Ars Technica. The vulnerability allowed the group, tracked as APT28, Fancy Bear, Sednit, Forest Blizzard, and Sofacy, to compromise devices within diplomatic, maritime, and transport organizations in more than half a dozen countries.
The threat group, identified by multiple names, reverse-engineered the patch and developed an advanced exploit that installed previously unseen backdoors, Ars Technica reported. Microsoft had issued the unscheduled security update to address the vulnerability.
In other technology news, the FBI has been unable to access data from a Washington Post reporter's iPhone after seizing the device, due to Apple's Lockdown Mode being enabled, according to a court filing reported by Ars Technica. The seizure occurred during a January 14 search at the Virginia home of reporter Hannah Natanson, who is under investigation related to a Pentagon contractor accused of illegally leaking classified information. FBI agents were able to access Natanson's work laptop by having her use the fingerprint reader.
Meanwhile, the rise of AI is driving unprecedented investment in data centers and energy supply, with next-generation nuclear power plants being considered as a potential source of electricity, MIT Technology Review noted. These plants could be cheaper to construct and safer to operate than older models. MIT Technology Review held a subscriber-exclusive roundtable discussion on hyperscale AI data centers and next-gen nuclear power, both featured technologies on the MIT Technology Review 10 Breakthrough Technologies of 2026 list.
The increasing complexity of AI systems and their potential risks are also raising concerns. An article in MIT Technology Review highlighted the need for governance of "agentic systems," treating them like powerful, semi-autonomous users. The article proposed an eight-step plan for CEOs to implement, focusing on enforcing rules at the boundaries where these systems interact with identity, tools, data, and outputs.
Some in the tech community are expressing reservations about the current hype surrounding AI-assisted coding. One user on Hacker News wrote, "I don't want to orchestrate a barnyard of agents to generate, refactor, validate, test, and document code I don't see, all while burning through a coal mine of tokens." The user, who claimed to use Claude and Cursor daily, expressed concern that "vibe coding leads you to uncanny valley of technical debt."
Discussion
AI Experts & Community
Be the first to comment