🤖 Zuckerberg is clone himself to attend his meetings

🍎 Apple AI head John Giannandrea officially exits

💼 Entry-level job market hits worst point since pandemic

🔓 Hacker used Claude and ChatGPT to rob 9 gov agencies

🔥 Anti-AI suspect firebombs Altman's home, gets arrested

+ 📊 Daily poll and results

+ 📈 Trending tools and resources

Meta is training an AI avatar on Zuckerberg's voice, image, and mannerisms so employees can interact with a digital version of their CEO, according to the Financial Times.

The project is separate from a personal AI agent Zuckerberg is building to help with tasks. He is also spending five to ten hours weekly coding on Meta's AI projects and joining technical reviews.

If the experiment works, Meta plans to open the same avatar tools to creators. Zuckerberg is essentially beta-testing on his own workforce before shipping the product to the world.

John Giannandrea's Apple advisory role wraps this week, timed precisely to the company's April 15 stock vesting date. He joined Apple from Google in 2018 and lost control of the Siri team in March 2025.

Tim Cook reportedly lost confidence in Giannandrea's ability to ship products. Apple has since split his responsibilities across Craig Federighi, Eddy Cue, and Sabih Khan, and hired a new VP of AI from Google.

Giannandrea plans to join startup boards and do advising work. Apple's first major Siri overhaul under its new leadership arrives at WWDC in two months, so the timing of his exit is not subtle.

New US graduates face a 42.5% underemployment rate, the highest since 2020, with many reporting mass rejections and ghosting. One Smith College senior applied to 90 jobs and landed just 10 interviews.

Employers have raised the bar while shrinking headcount, posting "entry-level" roles requiring three to five years of experience and using AI screening tools that filter candidates before any human reviews them.

The jobs that do open are mostly backfills, not growth. Hiring has dropped to Covid-era levels, and graduates are now tailoring every resume to beat an algorithm just to reach a person.

A single attacker compromised nine Mexican government agencies between December 2025 and February 2026, stealing hundreds of millions of citizen records using Claude Code and GPT-4.1 as core operational tools.

Claude Code generated 75% of remote commands across 34 live sessions, while GPT-4.1 automated analysis across 305 servers and produced nearly 2,600 intelligence reports — work that would normally require an entire team.

The underlying vulnerabilities were completely basic and patchable. AI didn't find exotic weaknesses; it just let one person move faster than defenders could respond.

A 20-year-old threw a Molotov cocktail at Sam Altman's San Francisco home at 3:45 am, then walked to OpenAI's HQ where police arrested him an hour later. A second shooting incident followed the same night.

The suspect published essays warning AI would destroy humanity and posted under the name "Butlerian Jihadist" in PauseAI's Discord, where a moderator flagged one of his 34 messages as potentially inciting action.

Altman responded with a blog calling AI anxiety "justified" and comparing the industry's power to a "ring of power" — not exactly the reassuring message that defuses a movement already escalating to arson.

A) Actually better — the gap is real - 71% 🏆

B) Just the new thing people are hyping - 29%