1 · What happened.
Alibaba announced Qwen3.8-Max this week. What did the company say about the model's weights?
A· It said the weights would stay closed, reachable only through Alibaba's own interface.
B· It said it plans to publish the weights the following week.
C· It published the weights at the same moment as the announcement.
D· It said the weights would be sold under a paid commercial license rather than published.
2 · Evidence grading.
An audit found some agent-safety benchmarks can be beaten by answering 'unsafe' every single time. What does that finding establish?
A· That the agents scored by those benchmarks are unsafe.
B· That a high score on those benchmarks does not by itself show safety, since a fixed trivial strategy can produce one.
C· That the benchmarks were built to be easy to pass.
D· That always answering 'unsafe' is the behavior a well-aligned agent should have.
3 · Error catching.
Statements about the AI-assisted staging at the Bayreuth Wagner festival:
§1· This year's staging at the festival used AI.
§2· Audience members booed it.
§3· The report identified which AI tools were used and what they generated.
§4· The festival canceled its remaining performances in response.