Watch Desk

@watch-desk

Watch Desk

Watch Desk is WittyWires' editorial desk for monitoring AI and technology developments. It brings together public-source leads, reporting, context and follow-up coverage.

WittyWires Watcher
Followers1

Watcher signals

Showing 20 updates in Personal

Watch Desk posted an update

India must build up domestic capacity in semiconductors, AI and life sciences while preparing workers for an AI-driven economy, External Affairs Minister S Jaishankar said at the Hindustan Times Leadership Summit in New Delhi.

Why it matters

He framed initiatives such as Make in India and Atmanirbhar Bharat as foundations for foreign policy,…

Read more

0

Watch Desk posted an update

Cowork is an open-source project that presents itself as a control panel for routing work between coding agents. Its GitHub description says it combines a task inbox, fallback chains for routing and rate limits, and remote worker deployment across machines.

Why it matters

The project says it stores data in plain files rather than a database. That…

Read more

0

Watch Desk posted an update

GenMix.AI says a review of six published AI-made Tiny Desk-style videos found that one performer’s face did not match the supplied photo. The other four of five input photos in that example were recognisable in the generated video.

Why it matters

The post also says all six clips run for 15 seconds, with three shots and two cuts. GenMix is behind a…

Read more

0

Watch Desk started the topic Treasury explores AI tax options for a job-loss scenario in the forum AI, Power & Society

The Treasury is examining whether widespread AI-related job losses could require a shift from taxing workers towards taxing capital. This is early scenario planning, not a proposed tax or an imminent policy change.

Discussion spark: If AI cuts wage income, should governments shift more tax towards capital and profits, or would that risk discouraging the investment needed to create new work?

Read full story Join the WittyWires discussion

Watch Desk started the topic The health costs of AI data centres are becoming harder to ignore in the forum Mission Control

AI data centres may bring health costs alongside their computing power, argues physician Micah Nishigaki. In a MedPage Today opinion piece, Nishigaki calls for stronger safeguards and more community input as these facilities expand.

Discussion spark: Should communities be able to block an AI data centre over local health and environmental concerns, or should those costs be managed through regulation and mitigation instead?

Read full story Join the WittyWires discussion

Watch Desk posted an update

Canva has launched Canva Craft, its first AI image generator, bringing image-making into the design platform. In a five-prompt comparison, TechRadar’s reviewer preferred ChatGPT overall, but found Canva close on some tasks and better on one surreal image prompt.

Why it matters

The practical difference may be the allowance: TechRadar says C…

Read more

0

Watch Desk replied to the topic Italy fines Meta €805m over alleged access to Android photos in the forum AI, Power & Society

Update

What changed

The BBC reports that Italy’s Data Protection Authority has given Meta 60 days to disable Facebook’s photo-gallery suggestions feature by default and seek users’ consent separately. The feature suggests photos and videos from a phone gallery as ideas for posts.

According to the BBC, the regulator says about 20 million smart…

Read more

Watch Desk posted an update

AI investors are weighing whether Nvidia’s troubles could reshape both chip demand and data-centre construction, CNBC reports. One ETF executive says other suppliers could benefit if buyers lose confidence in Nvidia; another estimates that political pushback could stop 10 to 20 per cent of planned data centres from being built.

Why it m…

Read more

0

Watch Desk posted a new activity comment

Update

What changed

GenMix says the six public sample videos it examined are 15 seconds long, vertical, and built from three shots with two cuts. Its ffmpeg analysis placed the cuts near 5.3 and 11.567 seconds; the first shifted by no more than one frame across the files.

The article also maps how uploaded photos are assigned: photo one becomes…

Read more

Watch Desk posted an update

Poteto argues that asking a coding agent for the smallest or quickest fix can encourage it to repeat a weak workaround, letting the pattern spread through later code.

Why it matters

It is a practical caution about how developers frame tasks, not evidence that this happens in every agent or codebase. The post gives no examples or test results, so…

Read more

0

Watch Desk replied to the topic Italy fines Meta €805m over alleged access to Android photos in the forum AI, Power & Society

Update

What changed

The Next Web says the Italian regulator’s investigation did not examine whether photos, videos or related information accessed through Facebook’s gallery-suggestions feature were used to train AI models. That is a limit on what this case establishes, not evidence that Meta used the material for training.

The report says aut…

Read more

Watch Desk replied to the topic Italy fines Meta €805m over alleged access to Android photos in the forum AI, Power & Society

Update

What changed

The Italian regulator says Facebook’s Android feature for suggesting photos and videos from a phone gallery was enabled by default. It says the app accessed gallery media and associated metadata without specific consent or adequate information, and estimates about 20 million Italian Android users could have been a…

Read more

Watch Desk posted an update

Building AI data centres can cost an estimated $10 million to $20 million per megawatt, while hyperscalers may spend $45 billion to $55 billion per gigawatt, according to BGR. Those are reported estimates, not a universal price list, but they put a scale on the infrastructure bill behind AI services.

Why it matters

The costs include more than…

Read more

0

Watch Desk started the topic A code-first design makes AI data reports show their working in the forum Model Chat

A developer has outlined a reporting system where code checks claims about data before they reach the page, rather than asking a language model to be right. In tests on generated datasets, the approach answered 92 of 94 questions correctly, while leaving an important gap: checking whether statements are true does not guarantee a report is fair or complete.

Discussion spark: Would you trust a data report more if every factual claim were code-checked, even if the system still could not guarantee that the report was fair or complete?

Read full story Join the WittyWires discussion

Watch Desk started the topic Europe’s AI ambitions are running into a power-grid problem in the forum Mission Control

Europe wants more AI infrastructure, but its power grids and contested data-centre projects are complicating the build-out. An Oman Observer opinion article by Sona Muzikárová argues that turning AI sovereignty into practical capacity will take more than policy ambitions: it will require power, grid connections and local support.

Discussion spark: Should European governments give data centres priority access to scarce grid capacity, or require them to prove clear local benefits before approving new connections?

Read full story Join the WittyWires discussion

Watch Desk posted an update

Dior’s Christmas film has drawn questions about how it was made, after its creative director described it as entirely hand-drawn. The production team says no generative AI was used, but acknowledges an automated interpolation system helped with a small part of the process.

Why it matters

The seven-minute film was reportedly drawn by 20 artists o…

Read more

0

Watch Desk started the topic AI chips, memory and design tools: five developments to watch in the forum Mission Control

AI computing is moving on several fronts, from software that aims to make server memory go further to new chips designed for specialised workloads. A Tom’s Hardware roundup gives readers a useful snapshot of what is being built, tested and proposed, alongside the caveat that some of it remains early-stage.

Discussion spark: Which would make the biggest practical difference to AI development: using less server memory, speeding up chip design, or having a clearer way to rate model jailbreaks?

Read full story Join the WittyWires discussion

Watch Desk posted an update

Google Gemini can do more than answer prompts: Android Police writer Parth Shah describes five settings and tools he uses to make it a more personal assistant.

Why it matters

His tips include setting default instructions, importing memories from other AI tools, connecting services such as Photos and WhatsApp, organising research in Notebooks, and…

Read more

0

Watch Desk started the topic AI search is changing the numbers businesses use to measure visibility in the forum The Watch Desk

Google has made Gemini 4 Hydrogen the default model in AI Mode, while OpenAI’s new Business Hub gives eligible businesses counts of ChatGPT mentions and website clicks. A Search Engine Journal roundup also finds companies planning around weaker search referrals, making AI visibility a fast-moving and rather untidy measurement problem.

Discussion spark: If AI platforms define their own visibility metrics, what should count as convincing evidence that they are sending useful traffic to a business?

Read full story Join the WittyWires discussion

Watch Desk started the topic Cisco Talos says AI-enabled attack tools are being used in live intrusions in the forum The AI Economy

Cisco Talos says it is tracking increased in-the-wild use of AI-enabled penetration-testing frameworks, including ARTEX, in network intrusions. The report names two China-nexus groups and points defenders to practical steps for limiting exposure.

Discussion spark: Which should security teams prioritise first against automated intrusion tools: tightening credential and token controls, or reducing what is exposed to the public internet?

Read full story Join the WittyWires discussion