March at Sony AI: Jesse Lai on writing The Principles of Diffusion Models, 11 papers accepted to ICASSP 2026, and Alice Xiang on the Me, Myself and AI podcast discussing FHIBE. Full recap: bit.ly/4c4oBxO #SonyAI #AIResearch #MachineLearning #ICASSP2026

SHIELD benchmarks how #vision-language models detect #AI-edited images without task-specific training. 👉 Read more: bit.ly/4dRuHE3

EnTruth offers a new way to trace unauthorized dataset use in #diffusionmodels with minimal image alteration. 👉 Full paper: bit.ly/4sXLlpO

StelLA shows how structured low-rank adaptation can outperform standard #LoRA without added compute. 👉 Read the paper: bit.ly/46PkK62

Music Arena brings live, human preference evaluation to text-to-music #research. 👉 Read more: bit.ly/4aMmeQH

FlashFoley enables real-time, controllable sketch-to-audio generation. 👉 Read the paper: bit.ly/46qQQVn

LADiBI solves blind inverse problems using latent diffusion—no training required. 👉 Full paper: bit.ly/46RbrCn

TalkCuts is a large-scale dataset for multi-shot human speech video generation. 👉 Full paper: t.co/AElsaEXUoB

January at Sony AI was about sharpening foundations: from scientific discovery and diffusion models to autonomy and learning. Catch our January highlights, including new #research perspectives, a diffusion models book, an #AAAI invited talk, and a look ahead to ICLR 2026.

Our 2025 Year in Review is here. This year we advanced responsible data practices with #FHIBE, introduced new tools for music and media creation, strengthened sensing and imaging pipelines, and expanded #RL research from #GTSophy to adaptive agents. Read the full recap: bit.ly/3YG9eoN

LLM-BRec = faster, more personal recs. 50% less training, 80% less inference, better results. 👉 Read the paper: bit.ly/45c7Q0K

Three new studies from #SonyAI explore fairness in music generation. From #unlearning and #attribution to #recognition and #protection. Creativity should begin, and remain, with people. Read more: bit.ly/4q9GhNK #AI #Music #Research

Smarter ensembles for translation. #SmartGen picks the best paths with RL — faster, leaner, better. 👉 Read the full paper: bit.ly/3YmJyNK

💫Proud to share that #FHIBE has been featured on the cover of @nature.com. The issue spotlights images from our globally diverse, consent-driven dataset designed to benchmark fairness in AI. Read the full feature via Nature: bit.ly/4pIpcu3

Alice Xiang, Global Head of AI Governance at Sony Group Corporation, and Lead Research Scientist at Sony AI, on FHIBE: Global diversity. True consent. Scientific rigor. Watch A Fair Reflection + explore fairnessbenchmark.ai.sony #FHIBE #FairAI #AIFairness #EthicalAI #SonyAI

Detect dataset use without insider access.
 SMI flags if your data trained an LLM or VLM.
 👉 Read the full research paper: bit.ly/48fZK9F

AI datasets shape how we’re seen. But who gets to decide what’s fair? Watch “A Fair Reflection,” our short film on the FHIBE dataset—built with consent, representation & mindfulness. 📽️ Visit the site to learn more: 👉 fairnessbenchmark.ai.sony

Built with consent, FHIBE is a new benchmark for fairness in vision tasks. 🎬 Our short film A Fair Reflection shows why that matters. Explore the film and benchmark at fairnessbenchmark.ai.sony #FHIBE #FairnessInAI #EthicalAI

What does it take to build a #dataset with #consent, diversity & transparency at its core? Discover FHIBE—the Fair Human-Centric Image Benchmark—and the insights from the team who built it: bit.ly/3LObF5r #SonyAI #FHIBE #EthicalAI #ResponsibleAI

Meet #FHIBE — just published in @nature.com. A Fair Human-Centric Image Benchmark built to reflect us all and expose bias in the data behind AI vision systems. Watch A Fair Reflection → fairnessbenchmark.ai.sony #AIEthics #FairAI #SonyAI

Sony AI’s FHIBE — just published in Nature — sets a new benchmark for evaluating fairness in AI models for computer vision. Hear from Michael Spranger in A Fair Reflection, our short film on FHIBE’s creation → fairnessbenchmark.ai.sony #FHIBE #EthicalAI #SonyAI

🚨 Big news: FHIBE is here. Sony AI’s Fair Human-Centric Image Benchmark, published in Nature, is the first globally diverse, consent-driven dataset to benchmark AI fairness. Explore the research + dataset: bit.ly/3JKaqnm #FHIBE #AIethics #SonyAI

A paper in Nature presents a database of more than 10,000 human images to evaluate biases in artificial intelligence models for human-centric computer vision, called the Fair Human-Centric Image Benchmark (FHIBE) and was developed by Sony AI. go.nature.com/4qP0MAK 🧪