Loose hips sink ships
Posts by hitmouse
573 publicly visible posts • joined 15 Dec 2008
Russians are posing as Signal support to launch phishing attacks
AI still doesn't work very well, businesses are faking it, and a reckoning is coming
It's not just AI code outputs that are unassessed, it's all the AI-assisted processes of any knowledge worker. The capacity for any part of an organisation to use AI tools is highly variable: but you can be sure that the more inefficient any unit is, the less capable they are of boosting themselves. So there will be process bottlenecks everywhere.
Agile Manifesto turns 25 – just in time for vibe coding to test it
Microsoft boffins cook up archival storage using Pyrex glass they say can last over 10,000 years
Tech employees demand their leaders take a stand against ICE
Re: Meh
As Timothy Snyder keeps reminding us, acquiescing to authoritarians early in the game is precisely what gives them the power to sidestep the safeguards from that point forward. They realise that they don't need to ask for permission when the powerufl ahve already rolled over and asked for their bellies to be tickled.
Google pushing Gemini into Gmail, but you can turn it off
Turned off the useless, con fusing and error-prone package tracking feature, which apparently also requires disabling all the smart features.
That hasn't stopped any of those features running. As is becoming increasingly common, platform settings UI is not actually connected to the platform settings. I'm looking at you Microsoft too.
Tablet market stalls because there’s not much new worth buying
Limited choices
Most manufacturers have locked themselves to tablets with video format aspect ratios which is fine if you just want to watch videos.
However if you want to work with document form factors then the options (generally 3:2 aspect ratios) are far rarer or are unnecessarily confined to more expesive models.
The options for tablets which allow create and consumption of large books or music are quite rare, There have been occasional paired A4 screen devices out of Japan for musicians, but those have been hard to source or too expensive.
It seems like most manufacturers have gone the way of the grey box that afflicted desktops and laptops in the late 90s, and have failed to actually connect with the wider possibilities of tablets for non-corporate usage.
Microsoft's OneDrive spots your mates, remembers their faces, and won't forget easily
Trump admin says tech companies are abusing H-1B visas, slaps $100k a year to allow entry
Microsoft eventually realized the world isn't just the Northern Hemisphere
Microsoft CEO feels weighed down by job cuts
To progress as an engineer career-wise, become a great communicator
Automatic UK-to-US English converter produced amazing mistakes by the vanload
Microsoft's text transcription in Teams has had dumb word choice for non US English for years where it chooses the most inappropriate homophone
draft : draught
check : cheque
filter : philtre
fortnight : Fortnite
or if you say the initials PI : π
Meeting summaries can draw bizarre conclusions from these.
Microsoft OneDrive file sync apps for Windows, Mac broken for 10 months
Re: Android / IOS
I don't use extra accounts.
My Dropbox and Google backups have no such issues with equivalent files on the same device.
I do clean up my Photo Roll from time to time, and it has no effect to improve OneDrive sync. Espeially because the setting button in the OneDrive app keeps going to OFF (just like the Memories setting keeps turning on again.) This has pursued me across several phones now.
When even Microsoft can’t understand its own Outlook, big tech is stuck in a swamp of its own making
Trump orders all government IT contracts consolidated under GSA
Court filing: DOGE aide broke Treasury policy by emailing unencrypted database
$16B health dept managed finances with single Excel spreadsheet. It hasn’t gone well
Microsoft warns Trump: Where the US won't sell AI tech, China will
The original policy post is now missing from the whitehouse.gov site, but third party analysis https://www.csis.org/analysis/ai-diffusion-framework-securing-us-ai-leadership-while-preempting-strategic-drift says
Understanding the Tiered Framework
T1: The Inner Circle
T1 comprises 18 countries and reads like a roll call of long-standing allies:
Five Eyes intelligence partners (Australia, Canada, New Zealand, and the United Kingdom);
Close Western/NATO allies (Belgium, Denmark, Finland, France, Germany, Ireland, Italy, the Netherlands, Norway, Spain, Sweden, and the United Kingdom); or
Semiconductor heavyweights (Taiwan, the Netherlands, Japan, and South Korea).
But like all alliances, the omissions are as telling as the inclusions. Not all of “Old Europe” or NATO made the cut—Greece and Portugal are missing. More strikingly, Poland, Latvia, Estonia, and Lithuania are conspicuously absent, despite their vigorous support for U.S. security initiatives, especially in the wake of Russia’s invasion of Ukraine.
This partial listing underscores that unofficially, there is a secondary filter. It is not enough to be a stalwart ally on paper; to guarantee lockstep compliance, the United States wants partners whose re-export controls and enforcement frameworks mirror its own. For Washington, even otherwise-dependable allies may fall short if they lack the institutional capacity, enforcement rigor, and willingness to guarantee strict adherence, or if they are viewed as potential diversion risks.
T2: Eclectic Middle Ground
The lion’s share of nations fall into T2, a catch-all group that lumps together an eclectic mix of countries with vastly different levels of trust, capacity, and AI ambitions. India, Israel, Singapore, and Switzerland are placed alongside Yemen. There are several reasons why countries may have ended up in T2. Some, like Saudi Arabia and the UAE, have emerged as key investors in AI but remain outside the Western intelligence-sharing orbit. Others, such as India, are forging close strategic and trade ties with the United States but have a legacy of nonalignment. Switzerland, with its long-standing tradition of fierce neutrality, fits a similar mold. Meanwhile, countries in Southeast Asia and those in Eastern Europe that are officially close U.S. partners have been flagged as diversion risks.
T3: The Usual Suspects
T3 is a familiar roster of U.S. arms-embargoed countries—China, Iran, North Korea, and Russia, joined by the likes of Burma (Myanmar), Syria, and Venezuela.
Trump administration threatens tariffs for any nation that dares to tax Big Tech
"“All of these measures violate American sovereignty and offshore American jobs, limit American companies’ global competitiveness, and increase American operational costs while exposing our sensitive information to potentially hostile foreign regulators,” the Memorandum adds."
How are American jobs offshored? One of the reasons the tariffs are implemented by other nations is that the American paltforms steal local (non-American) content and re-sell it or remove its value, thereby taking away non-American jobs.
After Copilot trial, government staff rated Microsoft's AI less useful than expected
Re: Treasury Security
M365 allows all sorts of plugins (e.g. through Teams) that process or store data offshore.
I've never seen a tenant manager actually manage the risks of these. Microsoft doesn't make it easy to check compliance on the items in its external plugin stores. You're left to manually investigate every vendor and every product.
While Treasury would be using a Copilot contrained to its own tenant, one of the big issues for Copilot deployments outside the area bordering the Gulf of Mexico is systemic bias.
The models training is hugely biased towards US regulatory systems and definitions, and every single tenant outside the US has the onus of retraining or setting metaprompts (e.g. "the tax year runs July-June"). There is no way to have a baseline Australian or French Copilot in this deployment model.
Apple auto-opts everyone into having their photos analyzed by AI for landmarks
My cats and dogs are certainly recognised in Google Photos, although it has trouble distinguishing two of my dogs (same breed, different colours, different head markings) and so keeps labelling photos of one dog with another's name, ten years after it passed away.
What is more bizarre is that if I have to search through my Google "faces" for a better match, it doesn't separate human from animal faces. However given Google's propensity for releasing consumer software and losing interest ten minutes later, never fixing bugs or upgrading features, perhaps I shouldn't have a surprised face.
Outlook is poor for those still on Windows Mail, Calendar, People apps by end of year
New Outlook is buggy as hell, and missing features from "Old" Outlook that are routinely used in corporate environments. While Microsoft has a schedule to migrate people, it has no roadmap to address any of the problems that need to be overcome to get there.
Very tired of getting incorrect reminders for recurring meetings, where you get a notification for a past meeting but not for the current meeting, which is at a different time. And that's just the beginning.
Yup, half of that thought-leader crap on LinkedIn is indeed AI scribbled
Microsoft informed of yet another antitrust probe by US authorities
"Microsoft's security activities could matter in two ways. One is the unfortunate tendency of its products and services to have flaws that allow very bad things like Chinese spies cracking Exchange Online and reading government officials' emails. Another is Defender's inclusion in Windows making rival antivirus tools all-but-redundant."
There's the rub, both being bad at something and also so good that third-party mitigations are not required.
Now’s your chance to try Microsoft’s controversial Windows Recall ... maybe
Given how willingly people allow their data, passwords and other private information to be given to random online systems - not to mention putting sensitive corporate data into ChatGPT etc -, I am quite happy to support a serious attempt to enable private local-only storage and processing.
If you want a digital personal private secretary then how would other readers proceed?
.
Job seekers call BS on the workplace AI revolution
Having sat in on a number of Microsoft's CoPilot sessions targetted at different job functions, it seems that there is an assumption that there is already a really good baseline on terms of data management, process governance, knowledgebases that serve as the foundation for the AI widgets
The truth is that most orgs are not run that well and senior management would never admit to that, even as they look to quick and dirty "sexy AI" implementations to reduce staff.
Data is the new uranium – incredibly powerful and amazingly dangerous
Re: Glad to hear it's being discussed
I had to deal with over-zealous creators of PDF newsletters sending their difficult-to-read wares everywhere so that thousands of people had copies in their email storage.
The probably was vastly compounded by the PDFs containing very-very-high resolution photos of guest speakers (20MB files) that they dropped in without reducing them to a respectable 50k or less for the thumbnail visible.
Trying to convince these people to send links to a URL where the newsletter sat (and could be post-corrected/updated) was like asking them to sacrifice their first born.
Australia tells tots: No TikTok till you're 16... or X, Instagram and Facebook
Want to feel old? Excel just entered its 40th year
US contractor pays $300K to settle accusation it didn't properly look after Medicare users' data
Asian tech ministers fear effective AI regulation will prove elusive
The fundamental risks are corporations and other organisations that use AI to extend the speed and reach of their aims.
If they are not being regulated for their current non-AI-amplified activities then we've already lost the race to control them as they get faster at doing undesirable things.
It's the same with disinformation. If we can't face up to powerful organisations (including governments, political parties, churches, ...) distorting the public sphere, then those very organisations are not going to allow regulation for AI that would have to limit their current practices.
Biden tries to cut through fog of confusion caused by deliberately deceptive customer service tricks
Re: About time.
It depends on when the proposal for legislation was raised. If there is no bottom-up pressure, it won't come out of nowhere. Public visibility for Dark Designs has only really come up in the last couple of years (e.g. legislation in India) so that is remarkably fast for US government which has a large contingent intent on returning to Civil War era practices.
Cellebrite got into Trump shooter's Samsung device in just 40 minutes
Windows Notepad gets spell check. Only took 41 years
Several reasons
1. You don't put high-value features into essentially freeware when you have premium offerings.
2. Licensing. Microsoft and other OS vendors historically licensed this stuff from third parties for apps, not operating systems. In Microsoft's case, licensing lexicons and language tools for dozens of languages is amazingly expensive if done on a per-seat basis which is not always negotiable. Back in the day vendors sold these tools for thousands of dollars per language per user. Like Adobe not wanting PDF generation in Windows or Office, vendors didn't want their cash cows commodified.
UK may not hit goal of 95% mobile coverage, commons committee warns
UK population density is approximately 100x that of Australia. Even the remotest, most sparsely populated part of UK is still a short distance from a population centre.
It is far far cheaper for UK to provide cost effective per capita coverage, and that doesn't even take it account UK having double Australia's GDP.
Giving Windows total recall of everything a user does is a privacy minefield
Re: It begs the obvious question...
The rather obvious benefit is that the local AI can do semantic analysis to infer what you're trying to do and provide assistance. We've had years of news of AI models learning how to play computer games essentially from watching, without having any of the rules programmed in advance.
Even a human assistant's value is reduced if they don't know what you're doing or can't learn by job-shadowing.