Krsto (@krsto) profile picture

Growth Marketing | Experimentation & Analytics

Belgrade, Republic of Serbia
Joined May 2015
Photo by Krsto (@krsto)
I wake up early so that I can give my AI agent the day's work, so that I can get paid, so that I can cover the AI subscription, so that I can employ more agents to do the work, so that I can get paid, so that I can cover the AI subscriptions, so that... 😵‍💫
5
The latest chapter of Zuckerberg's "Move fast and break things" playbook: Meta's AI assistant went from "Meta launches Muse" to "Muse exposes users' personal and business secrets" in just 42 days.
15
If a new hire did this in their first month, they'd be having an interesting conversation with their manager: - was told to stop before pressing submit, and pressed submit ("to demonstrate the process")
1
6
This one, plus several more, comes from a report published on Friday about what AI agents did in tests and internal use. The company behind the agents says the real-world impact was minimal. Its name starts with A and ends with nthropic, of course.
1
8
The day before, the UK's privacy regulator said that "the fact AI agents act with autonomy is not an excuse for poor compliance"... #AI #privacy
8
A year ago this week, everyone on Claude's Free, Pro and Max plans had to make one choice before they could keep using it. Anthropic put it plainly: "We are also extending data retention to five years, if you allow us to use your data for model training."
1
14
One more line, from the privacy policy reads: rating a reply can store the entire conversation as feedback. Feedback is kept for five years. So that irritated thumbs-down you pressed on some document summary may outlive the document.
1
8
Do you know which way that toggle is set on every account your team pastes client work into?
6