Daily Briefing — April 15, 2026

Someone threw a Molotov cocktail at Sam Altman's house. Then, two days later, it happened again. Around the same time, an Indianapolis city councilman had 13 shots fired at his front door after supporting a data center rezoning proposal. The alleged attacker in the Altman case was reportedly motivated by fears of human extinction tied to the AI race. That is not a fringe concern anymore, even if the method of expressing it absolutely is.

Here is the thing worth sitting with: the resistance to AI has been building for years across job displacement anxiety, climate concerns, and fears about unchecked development. Most of it has stayed in the territory of protests, hunger strikes, and strongly worded open letters. The groups pushing back against accelerated AI development were quick to condemn the violence. But the fact that it escalated at all tells you something about how high the emotional temperature has gotten around this technology.

For people working in or adjacent to the AI industry, this is a signal worth taking seriously. Because the underlying tension it reflects is real and it is not going away. The conversation about what AI does to jobs, communities, and the planet is getting louder. If your work touches AI in any way, ignoring that conversation is no longer a viable option.

Goldman Sachs CEO David Solomon went on record during an earnings call saying he is "hyper-aware" of the risks coming from Anthropic's newest AI model, Mythos. The short version: Anthropic itself warned that Mythos has an unprecedented ability to find and exploit flaws in IT systems, and Goldman, which already uses the Claude family of models internally, is now working closely with Anthropic and the US government to figure out what that actually means for their defenses.

Here is the part worth sitting with. When one of the largest financial institutions on the planet publicly admits it is actively monitoring a specific AI model as a security threat, that is not routine corporate caution. That is a signal. The same AI tools being rolled out to boost productivity across FinTech and banking are also, apparently, capable enough to give serious hackers a meaningful upgrade.

The uncomfortable reality is that most organizations are not Goldman Sachs. They do not have the budget, the government relationships, or the direct line to Anthropic to stay ahead of this. Your employer is probably somewhere on that spectrum between "vaguely aware" and "actively unprepared," and the gap between those two positions is closing faster than most security teams realize.

Google just added something called Skills to Chrome, which lets you save your favourite Gemini prompts and rerun them with a single click instead of typing or pasting the same thing over and over.

The bigger picture here is that Google is steadily turning Chrome into an AI operating layer, not just a place where you open tabs and forget about them. Gemini already controls parts of the browser interface, and now it is getting a shortcut system. This is the kind of slow, incremental entrenchment that does not feel dramatic until one day you realise your whole workflow runs through it.

What this means practically is that people who learn to build and organise good prompt libraries right now are going to be noticeably faster than people who are still winging it prompt by prompt. The gap between a thoughtful AI user and a casual one is starting to show up in actual output speed, not just quality.