AI News — Saturday, May 23, 2026

9
AI is being used to resurrect the voices of dead pilots

A new application of AI technology is enabling the resurrection of deceased pilots' voices, raising both ethical considerations and potential benefits for aviation training and historical preservation.

TechCrunchindustry
9
OpenAI named a Leader in enterprise coding agents by Gartner

Gartner has recognized OpenAI as a leading provider of enterprise coding agents, highlighting the company's significant impact and advanced capabilities in automated software development.

OpenAI Blogindustry
8
How VCs and founders use inflated ‘ARR’ to crown AI startups

An investigative report reveals how venture capitalists and startup founders may be inflating Annual Recurring Revenue (ARR) metrics to overvalue AI companies, potentially creating an unsustainable bubble.

TechCrunchindustry
8
Listen Labs raises $69M after viral billboard hiring stunt to scale AI customer interviews

Listen Labs secured $69 million in funding following a highly successful viral marketing campaign, aiming to expand its AI-powered platform for conducting and analyzing customer interviews at scale.

VentureBeatindustry
7
TransitLM: A Large-Scale Dataset and Benchmark for Map-Free Transit Route Generation

Researchers introduce TransitLM, a new large-scale dataset and benchmark designed to advance map-free transit route generation, enabling AI models to create efficient routes without relying on traditional map data.

Hugging Faceresearch
7
Perception or Prejudice: Can MLLMs Go Beyond First Impressions of Personality?

A new study investigates whether Multi-modal Large Language Models (MLLMs) can develop a deeper understanding of personality beyond initial superficial impressions, exploring the nuances of AI perception.

Hugging Faceresearch
7
How Virgin Atlantic ships faster with Codex

Virgin Atlantic is leveraging OpenAI's Codex to accelerate its software development and deployment processes, demonstrating a practical application of AI in streamlining enterprise operations.

OpenAI Blogproduct
7
The Most Concerning AI Risk of 2026

An article discusses the most pressing AI risk anticipated for 2026, prompting reflection on potential societal and technological challenges as AI continues to evolve rapidly.

Dev.toindustry
7
SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation

Researchers introduce SOLAR, a new autonomous agent designed for lifelong learning and continual adaptation through self-optimization, promising advancements in persistent AI systems.

arXivresearch
6
DelTA: Discriminative Token Credit Assignment for Reinforcement Learning from Verifiable Rewards

This paper presents DelTA, a novel method for discriminative token credit assignment that improves reinforcement learning by enabling models to learn more effectively from verifiable rewards.

Hugging Faceresearch
6
How we're using Gemini Embeddings to build a smarter, community-driven feed on DEV

DEV.to is enhancing its community-driven content feed by integrating Google's Gemini Embeddings, aiming to provide users with a more intelligent and personalized experience.

Dev.toproduct
6
Tool-Augmented Agent for Closed-loop Optimization,Simulation,and Modeling Orchestration

This paper proposes a tool-augmented agent capable of orchestrating closed-loop optimization, simulation, and modeling, enhancing the capabilities of AI agents in complex problem-solving scenarios.

arXivresearch
6
π-Bench: Evaluating Proactive Personal Assistant Agents in Long-Horizon Workflows

π-Bench is introduced as a new benchmark for thoroughly evaluating the performance of proactive personal assistant agents within complex, long-horizon workflows, addressing a critical need in agent development.

Hugging Faceresearch
6
Full Attention Strikes Back: Transferring Full Attention into Sparse within Hundred Training Steps

A new research paper demonstrates a method to efficiently transfer full attention mechanisms into sparse ones within a minimal number of training steps, potentially improving the efficiency of large models.

Hugging Faceresearch
6
ACC: Compiling Agent Trajectories for Long-Context Training

This research presents ACC, a novel approach for compiling agent trajectories to facilitate more effective long-context training, which is crucial for developing sophisticated AI agents.

Hugging Faceresearch