AI News: 28 Headlines No One Expected

AI News: 28 Headlines No One Expected

by Matt Wolfe

Applied Sciences & Engineering AI — News & Applications UYComputer scienceUYQArtificial intelligence (AI)
🎙 Matt Wolfe 👥 1.0M 📅 December 20, 2025 ⏱ 37 min 👁 89K 📄 news review 🧭 2026-08-28
Available in: English (current) Français

Keywords

GPT Image 1.5Flux 2 MaxSAM AudioGemini 3 FlashAI video editing

Summary

In this video, Matt Wolfe delivers a rapid-fire roundup of 28 AI news headlines from a particularly busy week in December 2025. He begins by briefly mentioning OpenAI’s GPT Image 1.5 and then tests Black Forest Labs’ Flux 2 Max, comparing its image editing capabilities to Nano Banana and GPT Image 1.5. He then demonstrates Meta’s SAM Audio model for isolating audio elements, followed by a sponsored segment for VibeCode. The video covers new video editing features from Adobe Firefly, Luma’s Ray 3 Modify, Kling’s motion control and lip-sync upgrades, and Alibaba’s Wan 2.6 model. The second half is a rapid-fire news roundup covering OpenAI’s app submissions, Google’s CC productivity agent, Gemini 2.5 text-to-speech, Gemini 3 Flash, GPT 5.2 Codex, new open models from NVIDIA, Xiaomi, and Manus, and the first AI model trained in space. He also mentions Microsoft’s Trellis 2 for image-to-3D, Amazon’s Alexa+ and Ring AI, Mistral OCR 3, Meta AI glasses updates, and the word ‘slop’ as Word of the Year. The video is packed with information, including hands-on tests and practical insights.

177 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides substantial value by offering hands-on testing of several newly released AI models, such as Flux 2 Max, SAM Audio, and Luma Ray 3, which goes beyond simply reading press releases. The argumentation is based on direct experience, with the creator showing real outputs and pointing out both strengths and weaknesses. For example, he demonstrates Flux 2 Max’s failure to follow complex instructions and notes the hallucination issues with Gemini 3 Flash. The rapid-fire format is efficient, covering a wide range of topics without deep dives, but the hands-on segments add practical value. The creator’s arguments are generally well-supported by the demonstrations, though some claims, like the Runway audio confusion, are left unresolved.

Scientific Rigor, Source Quality, Title Accuracy

The video demonstrates a good level of scientific rigor by providing links to official sources for most announcements, including OpenAI, Google, Meta, and NVIDIA. The creator distinguishes between his own testing and reported news, and he is transparent about uncertainties, such as the Runway audio feature. The title accurately reflects the content, as the video indeed covers 28 unexpected headlines. The sources are credible and directly linked in the description, enhancing the video’s reliability. However, the video is a news roundup, not a peer-reviewed analysis, so the rigor is appropriate for its format.

223 words

Title / Content Match

The title accurately reflects the content: a dense compilation of 28 AI news items, many of which were unexpected.

Quality & Reliability

7/10

The video is a rapid-fire news roundup with hands-on testing of several AI tools. The creator provides direct links to official sources for most announcements, but the testing is anecdotal and not peer-reviewed. The content is generally accurate but includes subjective impressions and occasional confusion (e.g., Runway audio).

Chapters

Cited Sources

Concurring Sources

  • OpenAI GPT Image 1.5 — Official announcement of GPT Image 1.5
  • Google Gemini 3 Flash — Official announcement of Gemini 3 Flash
  • Meta SAM Audio — Official Meta blog post on SAM Audio

Dissenting Sources

  • Runway Gen-4.5 Audio — The creator notes confusion about whether Runway Gen-4.5 generates audio, as his tests did not produce audio despite reports.

External References

Contribution & Novelties

The video provides a comprehensive and timely overview of a week’s worth of AI developments, with hands-on testing that adds practical insights beyond press releases. It highlights the rapid pace of innovation and the competitive landscape among major AI companies. The creator’s testing of image and video models offers a user perspective on their capabilities and limitations.

Pour aller plus loin :

  • Segment Anything Model (SAM) — Background on Meta’s SAM, which inspired SAM Audio.
  • Text-to-Speech Synthesis — Overview of TTS technology, relevant to Gemini 2.5 TTS.
  • Diffusion Models — Underlying technology for many image and video generation models mentioned.

100 words

Radar Profile

The radar profile shows high scores in quantity of information and technical level, reflecting the dense, hands-on nature of the video. Quality and reliability are moderate, as the content is based on personal testing and news reports rather than peer-reviewed research.

Reliability 7/10

💬 Positif. Sur les 30 commentaires analysés, la majorité exprime de l'appréciation pour le contenu et les tests, avec quelques suggestions d'amélioration (comme la lisibilité des effets visuels) et des demandes de sujets supplémentaires (robots, débats CGI).