AI Network: Crosswalk Hack, Talking Dolphins, OpenAI Developer Models, Meta Benchmark Cheating, NVIDIA US Factories, and ChatNode

In Today’s AI for Business News:

• Crosswalk buttons in California were hacked to play AI-generated voices mimicking Musk and Zuckerberg, raising public tech security concerns.
• Google teams up with researchers to decode dolphin communication using a new AI model trained on decades of underwater recordings.
• OpenAI rolls out new developer-only GPT-4.1 models with bigger context windows, better coding skills, and lower costs.
• Meta is called out for using a tuned version of its Llama 4 model to boost benchmark scores, sparking calls for more transparency.
• NVIDIA announces plans to manufacture AI supercomputers in the US, investing in new facilities and local chip production.

Prompt Tip – Ask AI to complete multiple goals in one prompt to get more tailored and time-saving results.

AI Tool to Try – ChatNode – build branded AI chatbots with content uploads, rules, and model choices – https://theaipreneur.pro/chatnode

Source:

https://www.tomshardware.com/tech-industry/cyber-security/crosswalks-in-silicon-valley-hacked-to-play-satirical-messages-from-musk-and-zuckerberg-sound-a-likes

https://blog.google/technology/ai/dolphingemma

https://openai.com/index/gpt-4-1

https://techcrunch.com/2025/04/11/metas-vanilla-maverick-ai-model-ranks-below-rivals-on-a-popular-chat-benchmark

https://blogs.nvidia.com/blog/nvidia-manufacture-american-made-ai-supercomputers-us