Skip to content
Tech News
clear
Topics: Today This Week This Month This Year

Adobe brings Firefly, Photoshop and 70+ tools into Slack via Slackbot and MCP app

Adobe announced that Slack users can now invoke apps like Firefly, Photoshop, Premiere, Acrobat and Illustrator directly through Slack's AI assistant, Slackbot, which interprets requests and calls the right Adobe tool while factoring in conversation context. Separately, more than 70 Adobe tools are being made available through a new Adobe for Slack MCP app. The integration launches first for Slack Business+ and Enterprise+ customers.

Undersea Fibre Cables Quietly Power Global Internet Traffic, Starting at Gibraltar

Most international internet traffic actually travels through physical fibre-optic cables buried beneath ocean floors, not through the air as commonly assumed. One such cable emerges from the sea near La Línea de la Concepción, Spain, crossing the Strait of Gibraltar to Ceuta, a Spanish enclave on Morocco's coast that until recently relied on a single ageing cable link to the mainland.

Entrepreneur op-ed argues AI should amplify, not replace, human creativity

An Entrepreneur contributor argues that AI, despite its ability to analyze emotion and language, cannot truly feel, sense danger in a room, or trust someone instinctively the way humans do. The piece contends AI's real value lies in handling information, research and execution while humans supply instinct, purpose and irrational confidence.

Study finds aging hippocampus blends separate memories rather than erasing them

Researchers scanning brains of adults aged 18 to 74 found that memory accuracy for face-object and face-scene pairings dropped sharply after young adulthood, with middle-aged and older participants performing similarly rather than declining gradually. In older adults, the same hippocampal activity patterns that supported accurate recall in younger people were instead linked to cross-category errors, such as confusing a scene with an object. The study, published in Cerebral Cortex, found that brain shrinkage and attention differences did not fully explain this shift.

Floating tree-covered island in BC's Williston Lake baffles officials, mistaken for AI hoax

A roughly two-and-a-half acre island covered in mature trees suddenly appeared in Williston Lake, a reservoir in northern British Columbia, prompting officials to initially suspect the viral footage was AI-generated before satellite imagery confirmed it was real. BC Hydro spokesman Bob Gammer said record-high water levels, the highest in 14 years, likely broke loose a mass of woody debris and vegetation from the shoreline, which then drifted into open water. The island has since continued to behave unpredictably, further puzzling researchers.

AI leaders gather in Chapel Hill as Bessent criticizes industry's public messaging

At the G20 Innovation Ministerial in Chapel Hill, North Carolina, top AI executives including Nvidia's Jensen Huang, Anthropic's Tom Brown and OpenAI's Sam Altman joined a session led by Commerce Secretary Howard Lutnick and OSTP Director Michael Kratsios. Treasury Secretary Scott Bessent separately said AI firms have done a poor job explaining themselves to the public, reflecting growing scrutiny of the sector and its data center footprint. Separately, Lutnick told CNBC the administration is drafting a framework for semiconductor tariffs.

Mistral AI enables user data training by default across Vibe and API, opt-out required

Mistral AI's consumer-facing Vibe product and its Studio/API services now use customer conversations, documents, and other input/output data for model training unless users actively opt out. Enterprise customers on Vibe are opted out by default, with admins controlling the toggle, while individual users must manually disable data sharing through account settings or an admin panel.

Developer builds B200 attention CUDA kernel reaching 94.4% of FlashAttention-4 speed

A technical blog post walks through building a custom attention kernel for Nvidia's Blackwell B200 GPU entirely from scratch, using CUDA and some PTX code. Through a 14-step progression of optimizations illustrated with 60 diagrams, the kernel reaches 94.4% of FlashAttention-4's performance on 4K, 8K, and 16K sequence shapes, and is demonstrated by plugging it into a video-generation model.

Today's top topics: openai google android authority elon musk nvidia anthropic ai agents apple techcrunch disrupt 2026 android
Browse all topics →  ·  Today's trending topics →