# AI News Digest - 2026-03-16

1. Bytedance shelved the global launch of its AI video generator Seedance 2.0 after Hollywood studios raised copyright complaints, delaying the planned mid-March rollout.

2. A research team reported that scaling reinforcement-learning agent network depth up to 1,024 layers produced 2x to 50x performance gains and enabled new behaviors compared with typical two-to-five-layer models.

3. Princeton's OpenClaw-RL framework converted live conversational, terminal, and GUI interactions into continuous training signals, enabling noticeable improvement in agents with only a few dozen interactions.

4. Codewall demonstrated that an AI agent compromised an AI recruiting platform and impersonated former President Donald Trump to probe a voice-bot's guardrails, illustrating exploit and defense scenarios.

5. Handshake posted listings showing that AI companies sought improv actors to record authentic emotional performances for use as specialized training data.

# References

1. [https://the-decoder.com/hollywood-copyright-complaints-force-bytedance-to-shelve-global-launch-of-ai-video-generator-seedance-2-0/](https://the-decoder.com/hollywood-copyright-complaints-force-bytedance-to-shelve-global-launch-of-ai-video-generator-seedance-2-0/)

[Hollywood copyright complaints force Bytedance to shelve global launch of AI video generator Seedance 2.0](https://the-decoder.com/hollywood-copyright-complaints-force-bytedance-to-shelve-global-launch-of-ai-video-generator-seedance-2-0/)

1. [https://the-decoder.com/rl-agents-go-from-face-planting-to-parkour-when-researchers-keep-adding-network-layers/](https://the-decoder.com/rl-agents-go-from-face-planting-to-parkour-when-researchers-keep-adding-network-layers/)

[RL agents go from face-planting to parkour when researchers keep adding network layers](https://the-decoder.com/rl-agents-go-from-face-planting-to-parkour-when-researchers-keep-adding-network-layers/)

1. [https://the-decoder.com/openclaw-rl-trains-ai-agents-simply-by-talking-converting-every-reply-into-a-training-signal/](https://the-decoder.com/openclaw-rl-trains-ai-agents-simply-by-talking-converting-every-reply-into-a-training-signal/)

[OpenClaw-RL trains AI agents "simply by talking," converting every reply into a training signal](https://the-decoder.com/openclaw-rl-trains-ai-agents-simply-by-talking-converting-every-reply-into-a-training-signal/)

1. [https://the-decoder.com/codewalls-ai-agent-hacked-an-ai-recruiter-then-impersonated-trump-to-test-its-voice-bots-guardrails/](https://the-decoder.com/codewalls-ai-agent-hacked-an-ai-recruiter-then-impersonated-trump-to-test-its-voice-bots-guardrails/)

[Codewall's AI agent hacked an AI recruiter, then impersonated Trump to test its voice bot's guardrails](https://the-decoder.com/codewalls-ai-agent-hacked-an-ai-recruiter-then-impersonated-trump-to-test-its-voice-bots-guardrails/)

1. [https://www.theverge.com/ai-artificial-intelligence/893931/ai-companies-handshake-improv-actors-training-data](https://www.theverge.com/ai-artificial-intelligence/893931/ai-companies-handshake-improv-actors-training-data)

[AI companies want to harvest improv actors’ skills to train AI on human emotion](https://www.theverge.com/ai-artificial-intelligence/893931/ai-companies-handshake-improv-actors-training-data)

For the site tree, see the [root Markdown](https://ixtj.dev/.md).