Insignary Launches Clarity AIR for Undeclared Open-Source Code Detection

Key Takeaways

• Insignary launched Clarity AIR to detect undeclared open-source and AI-generated code at the source-code level, producing AI Bills of Materials alongside traditional SBOMs.
• Orion Resource Partners raised approximately $2.2 billion for lithium battery development, backed by a $1.8 billion UAE-U.S. critical minerals initiative.
• Lithium Americas plans $1.3 to $1.6 billion in Phase 1 spending at Thacker Pass as AI reshapes the battery supply chain.
• EVE Energy achieved a 95% equipment effectiveness rate using AI, while AI-driven recycling reaches 95% recovery rates for lithium and cobalt.
• NVIDIA, KoBold Metals, and Atana Elements are among key players leveraging AI across the lithium battery value chain.
• Google DeepMind's AI hurricane model has been used by the National Hurricane Center since August 2020 to improve storm forecasting accuracy.
• Applied Digital's quarterly revenue surged to $341.9 million from $80.9 million due to AI data center sales, though losses grew with expansion costs.
• About 25 percent of employees using AI agents at work had to deactivate them due to inconsistent or risky outputs.
• Manatee County approved a 12-month moratorium on new data center applications over water supply and environmental concerns.
• a16z partner Anish Acharya says the tech industry focuses too much on AI regulation and not enough on the technology's practical potential.

Insignary launches Clarity AIR to find hidden open-source and AI code

Insignary released Clarity AIR, a tool that scans source code to find open-source and AI-written code not listed in manifests or SBOMs. It matches code against a fingerprint database and classifies AI-generated output line by line with a confidence score. The tool also creates an AI Bill of Materials alongside traditional open-source inventories. Clarity AIR is available now through Insignary and its partners, with trial licenses and a free demo offered.

Insignary launches Clarity AIR for undeclared code detection

Insignary introduced Clarity AIR to help security and compliance teams find code that developers never declared. The tool works at the source-code level instead of relying on manifests, detecting modified or AI-regenerated open-source code. It produces audit-ready SBOMs and reports after human review. Taek Wan Kim, CEO of Insignary, said verification requires examining the code itself, not just the manifest.

AI speeds up lithium battery development and draws big investment

BCC Research reports that AI is accelerating the lithium battery value chain from mining to manufacturing to recycling. Major investments include Orion Resource Partners raising about $2.2 billion and a $1.8 billion UAE-U.S. critical minerals initiative. EVE Energy achieved a 95% equipment effectiveness rate using AI, while companies like KoBold Metals and WeSort.AI use AI for exploration and recycling. Governments are also backing AI-assisted lithium projects to reduce import dependence.

AI reshapes lithium battery industry from mine to recycling

BCC Research highlights how AI is compressing development timelines and improving recovery rates across the lithium battery supply chain. Orion Resource Partners raised roughly $2.2 billion, and Lithium Americas plans $1.3 to $1.6 billion in Phase 1 spending at Thacker Pass. AI-driven recycling can reach 95% recovery rates for lithium and cobalt while cutting emissions by 60%. Key players include NVIDIA, KoBold Metals, EVE Energy, and Atana Elements.

China's AI startups match U.S. models but lack U.S. funding

China's AI models now capture a growing share of global token traffic, but funding lags far behind the U.S. Between 2023 and 2026, U.S. AI companies received over $380 billion in venture funding, while China received much less. Chinese startups face inflation, talent competition, and scarce venture capital, with state banks absorbing excess liquidity. Entrepreneurs are turning to Hong Kong listings, private credit, and customer revenue sharing to stay competitive.

Applied Digital AI Revenue Surges but Losses Grow

Applied Digital saw quarterly revenue jump to $341.9 million from $80.9 million as AI data center sales grew. However, costs rose to $404.3 million due to site expansion spending. The company holds $6.4 billion in debt against $3.7 billion in cash. Management plans to sign long-term deals with major hyperscalers to stabilize cash flows.

Jev Model Improves AI Decision Making in Chess and Search

Researchers studied the Jev probability model in two tasks involving decisions under tight constraints. In bullet chess, combining Jev with Stockfish search achieved above a 2200 Lichess rating. Using both Jev and Qwen3-32B as teachers improved results by 9.6 Elo over Qwen alone. In passage reranking, Jev labels trained a competitive reranker using just 21 minutes of API calls.

Employees Build AI Agents Without Company Approval

Many employees are creating their own AI agents at work without waiting for IT approval. About 25 percent of agent users had to deactivate agents due to inconsistent outputs. Seven percent of knowledge workers had to shut down agents that posed serious risks. IT leaders need to understand how these agents connect to company systems and data.

Manatee County Halts New AI Data Centers for 12 Months

Manatee County Commissioners approved a 12-month moratorium on new data center applications. The decision was driven by concerns over water supply and environmental impacts. Large data centers require substantial electricity and water, creating infrastructure demands similar to small municipalities. No data center applications are currently pending in the county.

New Method Studies How Language Agents Behave

Researchers proposed a framework for studying language agent behavior using trajectory abstractions. The method measures events at different levels and tests their stability across conditions. It constructs episode-level variables and connects every abstraction level back to original data. The approach aims to discover generalizable behavioral hypotheses across tasks and models.

a16z partner says AI regulation gets too much attention

Anish Acharya, a general partner at a16z, told CNBC that people talk too much about AI regulation. He believes the focus should be on what AI can do. He feels the technology's potential is being overlooked.

National Hurricane Center uses Google DeepMind AI to track Isaias

The National Hurricane Center in Miami used an AI hurricane model from Google DeepMind to forecast Hurricane Isaias. The center has used this model since August 2020. The AI uses machine learning to study satellite imagery, weather stations, and radar. It helped forecasters predict the storm's path, strength, and landfall areas more accurately.

Sources

• Insignary Launches Clarity AIR to Detect Undeclared Open-Source and AI-Written Code
• Insignary Launches Clarity AIR to Detect Undeclared Open-Source and AI-Written Code
• AI Integration Accelerates Across the Lithium Battery Value Chain, Unlocking New Investment Frontiers
• AI Integration Accelerates Across the Lithium Battery Value Chain, Unlocking New Investment Frontiers
• China’s AI startups can match the U.S.’s models. They can’t yet match the U.S.’s money
• Applied Digital’s AI Data Center Sales Jumped, But Losses Deepened
• From Probabilities to Decisions: Search and Multi-Teacher Distillation with Jev
• Your employees are building AI agents. Do you know what they’re doing?
• Manatee Commissioners put AI data centers on ice with 12-month moratorium
• Trajectory Abstraction for the Science of Language Agent Behavior
• AI regulation is ‘overdiscussed,’ says a16z partner
• Why the National Hurricane Center bet big on AI to forecast Isaias