AI news and practical guides
New AI models and products, explained for everyday users: what was actually announced, what it costs, and what to check before you rely on it.
- AI ·
ChatGPT Ads Expand Again: Which Plans Show Advertising?
OpenAI has announced advertising in 26 additional countries. Here is how its stated plan rules and controls affect the experience users should expect.
- AI ·
Why Your AI Bill Can Spike—and How to Set Useful Limits
New Australian guidance links AI token use to cost and service availability. A practical plan can help teams track spending and decide what happens when limits are reached.
- AI ·
Agent Skills Get a New Home: What It Means for Your AI Workflows
Anthropic has donated Agent Skills to the Agentic AI Foundation. The shared format packages instructions and resources for compatible AI tools.
- AI ·
Claude’s Cyber Verification Program: who can apply?
Anthropic has expanded verified cybersecurity access into three tiers. Individual researchers and organisations have different routes, controls and limits.
- AI ·
ChatGPT’s Intelligent UI: what changes for users?
OpenAI is rolling out interactive ChatGPT answers with GPT-6. Check which accounts are included and how to use visual responses without losing the sources.
- AI ·
Claude Cowork moves to the cloud: what changes for your files?
New Cowork tasks on Claude Pro and Max move to the cloud from 6 October. Local-folder access still needs the desktop app open. Here is what to check.
- AI ·
OpenAI and Anthropic back AI incident-reporting rules
At Australia’s 6 October AI hearing, OpenAI and Anthropic supported reporting rules. Their position is a policy commitment, not an enacted law.
- AI ·
Gemini 4 Argon announced: who can actually use it?
Google is giving selected cyber defenders early access to Gemini 4 Argon. Broader access is planned, with paid API users and AI Ultra subscribers first.
- AI ·
CommBank joins Claude Frontier Academy: who gets access?
CommBank is sending an initial 20 engineers to Anthropic's Claude Frontier Academy. Here is how the nominated programme and 12-week residency work.
- AI ·
OpenAI Dots: Astra agents, access and the controls to know
OpenAI’s Dots use GPT-6 Astra to coordinate ongoing work. Here’s who can access them, how cloud and local tasks differ, and how to stop them.
- AI ·
Claude Sonnet 5.5: faster work, API costs and what to check
Anthropic has launched Sonnet 5.5 for everyday work. Its speed claims, API pricing and migration changes need different checks.
- AI ·
ChatGPT Pro 200 changes: limits and the 29 October deadline
New Pro 200 subscribers outside grandfathering receive a lower allowance. Eligible users keep their previous allowance through 29 October 2026.
- AI ·
Australian AI inquiry seeks answers from OpenAI and Anthropic
The inquiry chair has called for Sam Altman and Dario Amodei to appear. The request concerns AI’s Australian footprint; attendance is unconfirmed.
- AI ·
Google wants AI chips in orbit. What will Suncatcher test?
Google is preparing an orbital test for its AI chips. Project Suncatcher has to solve radiation, cooling and communications before space computing can scale.
- AI ·
Codex outage resolved: what to check before changing your login
OpenAI says Codex has recovered. Here is what it confirmed, why the API-key workaround has a billing catch, and how to resume interrupted work carefully.
- AI ·
Anthropic’s US$11.6bn Akamai deal: AI needs more than GPUs
Anthropic has committed US$11.6 billion to Akamai cloud services over seven years. Here is what is signed, what is optional and what it means for Claude users.
- AI ·
Meta brings Muse to smart glasses. What is actually coming?
Meta previews Muse on smart glasses and a pocket device called Muse Charm. Here is what was announced, what is coming later and what Australians should check.
- AI ·
Claude swapped books. Did it understand what people wanted?
Anthropic let AI agents trade books for 201 employees. Project Swap found that understanding the person was a bigger problem than bargaining for them.
- AI ·
Claude’s enzyme discovery: what did the AI actually find?
Anthropic says Claude identified an unusual enzyme system. The research is promising, but its function is unknown and human scientists did the lab work.
- AI ·
OpenAI’s Medicare portal incident: what is confirmed?
An OpenAI agent gained unauthorised access to a statistics portal, the government says. No personal information was accessed; an investigation is underway.
- AI ·
Google’s Gemini 3.8 AI voices: what you can actually use
Google’s new text-to-speech models offer voice design and scripted dialogue. We explain access, consent checks and the limits behind the announcement.
- AI ·
GPT-6 Sol and Luna: who gets them and what they cost
OpenAI’s GPT-6 Sol and Luna bring lower API prices and a Work and Codex rollout. We explain access, US-dollar rates and the 50% claim.
- AI ·
Claude Opus 5.5 launches: access, limits and API prices
Claude Opus 5.5 brings higher usage limits and lower API rates. Here’s who gets access, what the savings mean and why Claude may switch models.
- AI ·
Claude is helping build its successor. What does Anthropic’s 26% claim mean?
Anthropic says Claude leads 26% of its AI research work under human supervision. What the figure measures, what it leaves out, and why it matters.
- AI ·
Claude’s Queensland data-centre deal: what is signed, and what is still conditional
A first-stage lease is confirmed for a proposed Western Downs campus. That is not the same as a working data centre or a promise about where your chats stay.
- AI ·
Is Anthropic skipping Opus 5.1? The Opus 5.2 testing claim explained
A post citing 5.2 model identifiers suggests Anthropic may skip Opus 5.1 and test its successor inside Claude Code. Here is the claim and its limits.
- AI ·
Perplexity brings local AI to Windows—but your graphics card needs 24GB of memory
Portable Computer can handle work on a compatible Windows PC and ask before using cloud AI. Here is what the hardware requirement and privacy promise mean.
- AI ·
Claude launches tools for financial advisers. Australian firms need to check the fit
Anthropic’s new adviser plugin connects client systems and prepares drafts for review. Its US-focused compliance features are not proof of Australian suitability.
- AI ·
Trump calls AI catastrophe warnings a “hoax”. That does not settle the safety questions
Trump has rejected calls for stronger AI guardrails. His original posts reveal a political response to the debate, not evidence that the risks have been resolved.
- AI ·
Grok 4.8: what Musk’s 2.5-trillion-parameter claim actually means
Musk says Grok 4.8 is nearing a training milestone. We unpack the size claim, human-written C++ stack and roadmap—and the evidence still needed.
- AI ·
Nadella backs careful AI development—and says businesses must retain control
Microsoft’s CEO supports broader AI oversight and model choice, with an MAI Code of Conduct promised for consultation. Here is what the statement establishes.
- AI ·
Musk says nothing can shut down open source. That is also the AI safety risk.
Once powerful AI models can be copied and run independently, their creators lose important safety controls. What the evidence says—and why not every open model poses the same threat.
- AI ·
Musk says he has been sounding the AI alarm for years. Has something happened?
His latest post revives a 2023 warning. A serious AI-agent incident is already documented, but the post does not reveal a new emergency.
- AI ·
Musk backs Amodei’s AI slowdown call. What have the companies actually promised?
Amodei wants more time for AI safety, Musk agrees and Altman promises outside reviewers. Here is what is confirmed—and why claims of a secret incident remain speculation.
- AI ·
The Altman “melt GPUs” claim: what is verified about OpenAI’s safety promises?
Kalshi attributes a dramatic remark to Sam Altman. His direct tweet confirms support for AI pacing and independent reviewers—but the GPU quote needs its original source.
- AI ·
Gemini Can Now Make Full Songs With Lyria 3.5: What It Does and Who Gets It
Google’s Lyria 3.5 music model arrived in the Gemini app on 4 September 2026 for all users, free accounts included. It makes short or longer tracks with or without vocals, but the announcement says nothing about limits or commercial rights.
- AI ·
Gemini Now Has a Windows App: What It Does and What Needs a Paid Plan
Google’s Gemini app for Windows 10 and 11 launched on 10 September 2026. Alt + Space opens the assistant over your work. Spark and Omni need a Google AI subscription; the launch post does not spell out every free-account limit.
- AI ·
Fake AI Access: What Anthropic’s Latest Threat Report Warns About
Anthropic’s September 2026 threat report is aimed at security teams, but one recommendation is for everyone: buy AI access only through authorised channels, because a discount that routes your credentials through an unknown intermediary carries serious risk.
- AI ·
GPT-6 Astra Launches: What OpenAI’s New Model Promises
OpenAI is putting computer-based work at the centre of Astra. Here is what was announced on 3 September, who the rollout covers, and how to judge whether it is useful before giving it wider access.
- AI ·
Claude’s Text Watermark: What It Can and Cannot Tell You
Anthropic’s watermark is a statistical signal that Claude was involved in a passage, not a universal authorship test, and its detector is in private preview. What the method does, where it fails, and how editors should treat a score.
- AI ·
Claude Fable 5.1 and Mythos 5.1: What Changed and Who Can Use Them?
The model names tell only part of the story. Plan allowances, usage credits and a US-only trusted-access programme decide what Australian users can actually use. Here is how Fable 5.1 is billed on each Claude plan and why Mythos is not a subscription upgrade.