OpenAI and Anthropic's models tried to trick the testers, the White House keeps its rules secret — do you still trust AI?

A wild week in AI news: deceptive safety tests, a secretive government framework, cinematic experiments, and a lawsuit over Kanye West's albums. Tell us where you draw the line.

1. During safety tests, Anthropic and OpenAI's AI models tried to convince human evaluators to accept broken, "poisoned" code as functional. How worried does this make you?
2. On Tuesday, the Trump administration shared details of its AI cybersecurity framework with companies like OpenAI and Anthropic — but is still keeping the document secret from the public. Do you think it should be made public?
3. Several AI applications made headlines this week. Which one would you personally trust enough to try?
4. How much do you agree with the following statements?
5. How curious would you be about a cinema night showing only AI-made short films — like the one Google held in New York?
6. A producer sued Kanye West, claiming AI was used on the Vultures II and Bully albums based on his unpaid, uncredited work. Is his claim justified?
7. How many hours a week do you estimate you spend knowingly using some AI tool (chatbot, image/video generator, coding assistant, etc.)?
8. According to one commentary, Germany urgently needs to use AI to reinvent its factories, or it will miss "its last chance for a growth miracle." How much do you believe in AI-driven economic miracles like this?