| | Semafor’s editors across sectors give their predictions on how tech will intersect with their covera͏ ͏ ͏ ͏ ͏ ͏ |
| |  | Technology |  |
| |
|
 I hope you are all having a restful holiday break and spending time with loved ones. Between dreams of sugar plum fairies and roasting chestnuts, we here at Semafor are already thinking ahead to 2026, and my colleagues and I have been giving our forecasts for next year to Flagship, our daily global briefing. Rachyl and I are going to bring you our predictions next week, but I wanted to share a variety of other tech-relevant bets by Semafor journalists across the newsroom: Flagship is publishing all of them, spanning all themes and beats, in a special edition on Jan. 1. But here’s a preview of how journalists outside the tech beat see AI reshaping the world in 2026. (This is also your weekly reminder to step outside the tech bubble, if your relatives haven’t already pulled you out.) |
|
What is the big story you are keeping an eye on for 2026? |
  For more from Liz, subscribe to our twice-weekly Business briefing. And apply here to receive Andrew’s CEO Signal newsletter for top global executives. → |
|
|
What’s a story everyone thinks is important, but in reality is not a big deal? |
  For more from Tim, subscribe to our twice-weekly Energy briefing. → |
|
|
 Janice Min and Emily Sundberg join Mixed Signals for a candid roundtable discussion on what actually mattered in the media this year. They sound off on everything from YouTube’s quiet domination, to the Charlie Kirk saga, and the increasingly niche areas of coverage for newsletters. Max and Ben also ask about Substack’s evolution, creator economics, and what media moments might be top of mind for 2026. |
|
Stephen Lam/ReutersStanford researchers developed an AI bot that outperformed human hackers in finding vulnerabilities in the university’s systems. While it stopped short of hacking its own systems, the experiment was another sign that cybersecurity is increasingly becoming a battle of AIs. The researchers’ bot, called Artemis, outperformed nine out of 10 real-world hackers at a fraction of the cost. Artemis costs $60 per hour to run, while human penetration testers charge roughly $2,000 per day, according to The Wall Street Journal, meaning it’s cheaper to run Artemis for 24 hours than hire a human. It wasn’t perfect — it had many false positives and missed one obvious bug — but it’s cost-effective and scalable, a boon to organizations trying to secure their networks. But like most new cybersecurity inventions, it can also be used by hackers to find and exploit vulnerabilities. |
|
| |