Google is said to be preparing to release Gemini 3.8 Flash on Thursday, according to the report cited in the source article, as competition in frontier AI models intensifies following the arrival of Claude 5.1 and ahead of OpenAI’s next Astra reveal. The same report says Google has scrapped Gemini 3.5 Pro and shifted attention to Gemini 4.0, which is described as performing well in pre-training evaluation while post-training work continues. Ahead of that reported launch, Google announced a new agentic video understanding capability for Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite through Google AI Studio and the Gemini Enterprise Agent Platform. In a blog post, Google said enabling the feature on standard video analysis benchmarks can cut token use by as much as 88%, lower analysis costs by as much as 66%, and improve accuracy by as much as 7%, while keeping standard API token pricing. The article frames the update as part of a broader race across Google, OpenAI, Anthropic, and xAI-related efforts, with coding performance, video understanding, and agent capabilities emerging as key battlegrounds.

OpenAI has released the GitHub repository for Codex Security under the Apache-2.0 license, but the open-source code covers only the tooling layer used to trigger scans, manage findings and integrate with CI. The underlying service that reads code, reproduces vulnerabilities in isolated environments and generates patches still runs on OpenAI’s own servers, so access still requires an OpenAI login and Codex Security permissions. According to the documentation cited in the source article, Codex Security remains in research preview and is available to ChatGPT Pro, Business, Edu and Enterprise users, with some enterprise accounts also needing admin approval. The article walks through environment requirements, install commands, dry-run validation, model configuration, SARIF export and CI usage, then highlights five common pitfalls: max-cost not acting as a hard cap, output directories being blocked if placed inside a repository, API keys taking precedence over ChatGPT login in non-interactive contexts, Python 3.10 needing tomli, and MCP being limited to read-only metadata rather than live scans. It also covers local scan history in SQLite, false-positive handling, scan comparison, batch scanning across GitHub repositories, and the related TypeScript SDK APIs.

A joint report from SlowMist and Bitget maps the main security threats facing AI agents in Web3 trading, including prompt injection, malicious skill supply chains, API key abuse, sensitive data leaks, and irreversible on-chain execution.

PI token jumps over 30% in a day driven by Kraken listing and mainnet upgrade. Key resistance at $0.30 as community anticipates Pi Day announcements.

OKX launches X Drops with 500,000 PI tokens reward pool a day before Pi2Day. PI coin price drops to $0.1265, down 96% from ATH. Binance listing remains uncertain pending V26 mainnet upgrade.

Pi Network’s PI token has bounced 23% from this week’s low to $0.1870, lifting market cap above $1.5 billion. The move tracked a broader crypto recovery, but token unlocks, limited exchange listings, and weak technicals still weigh on the outlook.

Kraken announced listing of Pi Network's PI token, triggering a 30% surge and 300% volume spike. Bybit's CEO had rejected PI earlier citing a Chinese police fraud alert, reigniting debate.

Pi Network dropped to around $0.1450 after broader crypto weakness, heavier selling, and continued token unlocks weighed on price. Technical patterns in the source also point to a still-fragile setup.