Hashtag Trending

Ukraine's strikes disrupted three Yandex data centres—but did they also silence Russian propaganda bots? The damage is documented. The social-media claims remain unproven.
On Hashtag Trending for Wednesday, October 14, 2026, host Jim Love examines the service outages and the evidence behind claims of declining political bot activity. Cloudflare's measurements show automated web requests, including search crawlers—not a count of propaganda accounts.
Also today: Anthropic offers free AI security scans for eligible open-source projects through OSS Scanner, with reports delivered without human review. Scammers exploit calendar invitations to deliver malicious links, sometimes from hacked accounts belonging to people you trust. And a CNN investigation documents threats and arrests amid America's data-centre backlash, raising questions about community consent and the consequences of threatening language.
CHAPTERS
00:00 Today's Headlines
00:41 Yandex Strikes Fallout
01:38 Did Bots Go Quiet
04:15 Anthropic OSS Scanner
06:24 Calendar Invite Scams
08:56 Data Center Backlash
10:47 Community Conflict Costs
13:16 Closing Thoughts

AI has fewer safety rules than a washing machine, says Microsoft President Brad Smith. Jim Love examines his call for global AI safeguards, the growing data centre backlash, Yandex's traffic collapse after Ukrainian strikes, and a Canadian developer's challenge to police surveillance.
On Hashtag Trending for Tuesday, October 13, 2026:
• AI safety and oversight: Smith calls for incident reporting, notification of victims and an emergency brake for AI systems. We compare those proposals with the White House's voluntary safety agreement and independent audit commitments, and discuss what the OpenAI–Hugging Face incident reveals about the difficulty of investigating AI failures.
• Data centre backlash: Bill Gates argues that blocking local construction won't stop AI because facilities will be built elsewhere. Canadian and U.S. surveys show substantial opposition. Finland pauses work at two Google sites, while the shift from Singapore to Malaysia illustrates how resource pressures can move across borders. Federal Reserve officials also identify AI investment as a contributor to inflation pressures.
• Yandex outages: Automated traffic from a Yandex network approached zero after Ukrainian strikes damaged three Russian data centres. The measurements include ordinary search-engine crawlers and do not establish a collapse in Russian propaganda accounts. Taxi and delivery outages beyond Russia raise questions about cross-border service dependencies, even when personal data is stored locally.
• Turning surveillance around: Canadian developer Anthony Sistilli builds a Flock-style camera to identify police vehicles, then receives a visit from Peel Regional Police. We examine officers' concerns and documented misuse of vehicle-tracking systems in the United States.
CHAPTERS
00:00 Today's Headlines
00:25 AI Safety Rules Debate
02:02 OpenAI Incident Lessons
03:10 Data Center Backlash
05:33 Yandex Outages After Strikes
07:38 Turning Cameras on Police
09:06 Wrap Up and Support
Subscribe to Hashtag Trending for technology news with Jim Love. Share the show and leave your thoughts in the comments.
Contact Jim or support the show:
https://technewsday.ca
https://technewsday.com
Use the Contact Us page to get in touch, or visit the Donate tab to support the podcast.

Thus Spoke Compute: Is AI a "Normal Technology," or Something Much Bigger?
The hosts discuss an AI-generated piece called "Thus Spoke Compute," framed as a rebuttal to the idea that AI will diffuse slowly like electricity or the telephone, and use it to explore claims of accelerating, self-compounding progress (agents, robots building robots, parallel "million minds"). They segue to a major story about OpenAI releasing 722 math-related manuscripts and summaries from a larger set of problems, reportedly using an internal model and Lean for formalization, and debate the impact on mathematicians, human purpose, and identity as more elite work becomes replaceable. The conversation covers resistance to AI in medicine and jobs, alignment and control limits, agent "swarms," sandbox escape concerns, and safety debates including Anthropic's anti-abuse policy. They also note a new ChatGPT "intelligent UI," Grok's agentic setup, and France's new trillion-parameter Mistral model.
00:00 Is AI Just Normal Tech
01:25 Thus Spoke Compute
01:56 Doomer Takes Get Roasted
03:27 Swarm Solves Hard Math
04:53 Project Synapse Intro
09:36 Dyson Sphere Reveal
10:29 OpenAI Breaks Math
13:37 Jobs Identity Crisis
25:39 UBI And AI Medicine
31:43 Token Predictors Like Us
35:51 Training Control Problems
37:46 Agents The Swarm Story
39:43 Sandbox Escape Security
40:28 Agents Already Escaped
41:32 AI Is Not Monolithic
42:21 Spam Filters And Uncertainty
44:33 Can We Verify Outputs
46:37 Safety Leaders Under Fire
49:03 Testing GrokBot Agents
52:47 Why Grok Feels Uncontrolled
57:05 Control And Alignment Limits
01:00:24 GPT-6 Intelligent UI
01:05:10 Live Demo Of New UI
01:10:34 Daily Briefings And Agents
01:13:26 Europe Builds Sovereign AI
01:15:39 Be Nice To Claude
01:17:53 Touch Grass And Wrap Up

Could AI-built software challenge Adobe's subscription business? An experienced developer has released free, open-source alternatives to seven Adobe applications—but replacing mature professional tools is a much bigger job than launching a project.
Jim Love examines ArtCraft, the limits of an AI-training copyright ruling, SpaceX's proposed $40 billion NVIDIA financing, Canada's controversial lawful access bill, and parents using AI to avoid expensive school photo packages.
ADOBE ALTERNATIVES AND THE SAASPOCALYPSE
Brandon Thomas says he has fifteen years of industry experience, including about a decade with Rust, and a funded team to support the work. His feature-matching timeline has shifted, and PhotoCraft's documentation acknowledges that it cannot yet replace Photoshop for daily professional work. Could reliable everyday features still be enough for many customers? We also discuss CHISEL's AI-assisted reconstruction of source code from compiled functions, and why that is different from recreating an entire commercial application.
AI TRAINING AND FAIR USE
A US appeals court rejects ROSS Intelligence's fair use defence for training a competing legal-search service using Westlaw's copyrighted editorial headnotes. We explain the four fair use factors and the ruling's limits: the court explicitly distinguished this system from generative AI. The decision does not settle every AI-training copyright dispute.
SPACEX'S PROPOSED $40 BILLION NVIDIA DEAL
SpaceX is reportedly seeking a financing package for NVIDIA AI hardware. We examine the difference between hardware's working life and profitable life, the repayment obligations that come with debt, and the company's existing borrowings and reported losses. The financing is proposed, not completed; the processor count discussed is an estimate, not a confirmed order.
CANADA'S LAWFUL ACCESS BILL
Toronto-based Psiphon is preparing to move its operations if Bill C-22 passes in its current form. Its censorship-circumvention software serves around twenty million monthly users. Ottawa says amendments protect end-to-end encryption and guard against systemic vulnerabilities. Psiphon remains concerned about user safety and the obligations companies could face. The bill is before the Senate and is not yet law.
AI SCHOOL PHOTOS
Parents are using their own photographs and AI-generated studio backgrounds instead of traditional school photo packages. Reported examples include more than $100 for photos of three children and nearly $170 for four digital files. These are individual reports, not average prices—and some parents say AI changed their children's faces.
Subscribe to Hashtag Trending for technology news with Jim Love. What would make you switch from a subscription app to an open-source alternative? Tell us in the comments.
Contact Jim or support the show:
https://technewsday.ca
https://technewsday.com
Use the Contact Us page, or visit the Donate tab to support the show.
SOURCES AND FURTHER READING
Adobe alternatives:
https://www.techspot.com/news/114136-software-over-developer-uses-ai-build-free-adobe.html
https://github.com/storytold/photocraft
CHISEL research:
https://arxiv.org/abs/2608.27981
AI training copyright case:
https://www.techspot.com/news/114058-court-ruled-training-ai-someone-else-editorial-work.html
SpaceX financing:
https://www.reuters.com/business/media-telecom/spacex-seeks-40-billion-buy-nvidia-chips-ft-reports-2026-10-06/
Canada and Psiphon:
https://www.theglobeandmail.com/politics/article-psiphon-vpn-provider-plans-to-quit-canada-over-lawful-access-bill/
https://www.parl.ca/DocumentViewer/en/45-1/bill/C-22/third-reading
School photos:
https://www.dexerto.com/entertainment/parents-are-using-ai-to-generate-pictures-of-their-kids-to-save-money-on-school-photos-3416112/
 
00:00 Today's Tech Headlines
00:33 Has the SaaSPocalypse Finally Arrived?
02:35 AI Training: Appeals Court Rejects Fair Use Defence
04:50 SpaceX's $40 Billion NVIDIA Financing Plan
07:28 Could Canada's Bill C-22 Drive Psiphon Out?
09:59 Parents Use AI for School Photos
12:12 Wrap-Up and Support the Show

OpenAI has published 722 manuscripts containing AI-generated mathematical results. The release raises a practical question: how will researchers verify the claims and work out what these results actually establish?
On Hashtag Trending for Thursday, October 8, 2026, Jim Love covers OpenAI's math research, ChatGPT's new Intelligent UI, Mistral's European AI model and the continuing VMware customer backlash.
OPENAI'S MATH RESULTS
The 722 manuscripts fall into 372 families of related results, including partial advances and claimed solutions. They do not represent 722 separate problems solved. We discuss claimed progress toward the Riemann hypothesis, computer-checkable proofs in Lean and the work still facing mathematical experts.
CHATGPT'S INTELLIGENT UI
OpenAI demonstrates interactive answers that let you explore a bicycle's components, build an app to split a dinner bill and adjust a complete dinner plan. We also cover GPT-6's response speed, OpenAI's safety and accuracy claims, and the rollout across paid and free plans.
See the demonstrations: https://openai.com/index/gpt-6-for-everyone/
MISTRAL LARGE 4 — "LE CHONK"
Paris-based Mistral AI introduces a multimodal, mixture-of-experts model with just over a trillion parameters and a listed million-token context window. Trained from scratch in European data centres, it offers another option for organizations seeking more control over their AI. We examine the benchmark strengths and limitations. The public preview is available, with downloadable weights promised by the end of October; the final licence terms still need checking.
VMWARE CUSTOMERS LOOK FOR ALTERNATIVES
Following Broadcom's licensing and support changes, customers have reported cost increases of up to 1,000 percent. A survey of 269 VMware customer representatives finds 90 percent considering alternatives because of higher licensing costs. We discuss the difficulties of migration and the survey's sponsor, third-party support provider Rimini Street.
CHAPTERS
00:00 Headlines and Intro
00:30 OpenAI's 722 Math Manuscripts
02:14 What AI Math Could Mean for Science
03:12 ChatGPT Intelligent UI Demonstrations
04:42 GPT-6 Speed, Safety and Rollout
06:06 Mistral Large 4: "Le Chonk"
07:39 Benchmarks, Open Weights and AI Sovereignty
11:30 VMware Customers Seek an Exit
13:32 Survey Findings and Migration Barriers
15:13 Wrap-Up and How to Support the Show
Subscribe to Hashtag Trending for technology news with Jim Love. Share your thoughts in the comments.
Contact Jim or support the show:
https://technewsday.ca