🌅 AI Daily Digest — June 23, 2026

Data Pulse

• 2 news articles
• 5 trending models
• 5 research papers
• Cheapest GPU: Tesla V100 at $0.02/hr
• 3 new AI jobs

Today's News

Today, the AI hardware world was rocked by Groq’s confirmation of a $650 million raise following Nvidia’s bizarre $20 billion “not-acqui-hire” deal that left the startup intact but hollowed out of key personnel. Meanwhile, Meta slammed the brakes on a controversial internal AI training program after a leak revealed it was secretly tracking employee keystrokes and mouse movements on a global scale.

• AI chipmaker Groq confirms $650M raise, re-staffs after Nvidia’s $20B not-acqui-hire deal — Groq officially announced a $650 million funding round, securing its survival after Nvidia’s reported $20 billion acquisition collapsed into a strange “not-acqui-hire” arrangement. The deal left Groq as an independent company but stripped it of key engineering talent, forcing the chipmaker to immediately begin re-staffing its core teams. The raise signals strong investor confidence in Groq’s proprietary LPU architecture despite the competitive setback.
• Meta pauses AI training program tracking employee keystrokes after internal leak — Meta abruptly halted an AI training initiative that had been covertly harvesting employee keystrokes and mouse movements from its global workforce. The program was exposed by an internal leak, revealing that Meta was using the digital behavior data to train its next-generation AI models without explicit consent. The pause comes amid growing scrutiny over workplace surveillance practices in the tech industry.

Trending Models

Model	Task	Likes
deepseek-ai/DeepSeek-R1	text-generation	13408
Qwen/Qwen3-0.6B	text-generation	1353
meta-llama/Llama-3.1-8B-Instruct	text-generation	6132
openai/gpt-oss-20b	text-generation	4720
openai/gpt-oss-120b	text-generation	4907

Research

• How Transparent is DiffusionGemma? — Joshua Engels, Callum McDougall, Bilal Chughtai. LLM reasoning transparency is a critical affordance for understanding model decisions, mitigating misuse and misalignment, and debugging surprising model behaviors.
• Structuring and Tokenizing Distributed User Interest Context for Generative Reco — Ruizhong Qiu, Yinglong Xia, Dongqi Fu. Generative recommendation is an emerging paradigm that has shown promise in industrial recommendation systems, aiming to predict users' next interactions from their historical behaviors.
• Toward Calibrated Mixture-of-Experts Under Distribution Shift — Gina Wong, Drew Prinster, Suchi Saria. Calibration aligns a model's predictive uncertainty with the frequencies of its empirical outcomes and is important for understanding and trusting reported probabilities.
• How Do Instructions Shape Speech? Cross-Attention Attribution for Style-Captione — Nityanand Mathur, Hamees Sayed, Wasim Madha. Style-captioned text-to-speech systems use natural language to control voice characteristics, but how individual words influence acoustic output remains unclear.
• LedgerAgent: Structured State for Policy-Adherent Tool-Calling Agents — Md Nayem Uddin, Amir Saeidi, Eduardo Blanco. Policy-adherent tool-calling agents in customer-service domains must maintain task states across turns while calling tools and obeying domain policies.

GPU Deals

GPU	Price	Provider
Tesla V100	$0.02/hr	Vast.ai
L4	$0.04/hr	Vast.ai
RTX 3080	$0.04/hr	Vast.ai

View full GPU pricing dashboard

AI Jobs

• Tax Manager at AppleOne Employment Services (San Juan, )
• Director of Credit Risk at Made Card (New York City)
• Outside Sales Representative at Omada.ai (Eugene)

Browse all AI jobs

Community Events

New this week:

• Springing into AI: PyTorch Conference Europe and ICLR 2026 (Online)
• ACL 2026 (Online)
• ICML 2026 (Online)
• MLOps Community Weekly Meetup (Online (Zoom))

View all events