Character.AI enters microdrama with AI-animated video series

Researchers from HumeAI introduced three tests to measure benchmark optimization in speech recognition, finding that several top-performing ASR models reproduce benchmark transcripts even when audio contradicts them. Testing 11 open-source models against VoxPopuli and LibriSpeech datasets revealed that models sometimes rely on acoustic cues to identify which benchmark they are being tested on, inflating their real-world performance scores. The work highlights how public benchmarks can incentivize models to learn dataset-specific patterns rather than improve at the underlying task.

Wispr raised $280 million in new funding at a $2 billion valuation, bringing its total funding to over $361 million. The funding round signals investor confidence in the voice AI company as it expands beyond dictation use cases. The company is positioning itself for growth in a competitive market for speech recognition and voice-based AI applications.

LTX released LTX-2.5, an open-weights video generation model that produces 10-second clips in 6.8 seconds on Nvidia GB200 chips, with native multishot support and improved quality. The model is available free for organizations under $10 million ARR on Hugging Face, ComfyUI, and via API. LTX claims 33 million downloads across its model family and reports a 67% win rate in blind quality tests against competing models.

Ford is launching an AI-powered chatbot assistant in its Ford and Lincoln mobile apps that can answer questions about vehicle capabilities, fuel levels, cargo capacity, and towing specifications. The assistant is linked to individual customer vehicles and can provide information relevant to specific makes and models. Ford plans to expand the tool to include a voice-powered version.