The short version
Key points
- Australian businesses and organisations using branded SMS sender IDs need to register those names from 1 July 2026.
- Unregistered branded messages may appear as unverified and be grouped into one thread with other unverified messages.
- An unverified label is a warning sign, but it does not automatically prove that a message is a scam.
- Anthropic said it suspended access globally after a US directive applied to foreign nationals, including employees outside and inside the US.
- The public disagreement centres on whether the models were uniquely dangerous or could perform security-related tasks that other AI models can also perform.
Australia’s branded SMS change
A sender ID is the brand name displayed at the top of an SMS instead of a mobile number. It can be used by businesses and organisations such as banks, government agencies, delivery services, healthcare providers and other companies.
From 1 July 2026, Australian businesses and organisations sending messages under a branded sender ID need to have that sender ID registered. If it has not been registered, the message may be shown as unverified. Unverified messages will also be grouped together in a single thread.
The change is designed to make it harder for scammers to impersonate trusted brands. SMS sender names have been easy to abuse, allowing a fraudulent message to appear as though it came from a bank, government agency, delivery company or another familiar organisation. The problem is not only the link in a scam message; it is the borrowed trust attached to the sender name.
What unverified messages mean for users
An unverified label should be treated as a reason to slow down, particularly when the message creates urgency. Common scam themes mentioned in the report include failed toll payments, delayed parcels, account closures and supposed refunds.
Users should not click links, reply with personal information or send money in response to an unexpected message. Instead, they should visit the organisation’s official website, use its official app, or call a number found on a genuine bill or website. Contact details and links supplied in the SMS should not be used for verification.
The change will not stop every scam. It is intended to add a visible warning that may interrupt the pressure and urgency scammers rely on. An unverified message is not automatically fraudulent, because a legitimate business may not have registered its sender ID, but it should not be trusted solely because it contains a familiar brand name or personal details.
The impact on businesses
Businesses that use branded SMS messages may find that unregistered messages look less trustworthy after the change. Customers could ignore appointment reminders, bills or delivery updates, or contact support to check whether a message is genuine.
Businesses using a brand name in the position where a phone number would normally appear should speak with their telecommunications provider or SMS provider about registering the sender ID. The change is less relevant to messages sent from an ordinary phone number rather than a branded name.
Anthropic’s Mythos 5 and Fable 5 models
The second story concerns Anthropic, the company behind Claude. Fable 5 was reportedly released publicly around 10 June 2026, while Mythos 5 was more restricted and made available to selected partners through Project Glass Wing. The report describes the models as particularly capable in coding and security-related work, with additional safety layers on Fable 5.
On 12 June, Anthropic said the US government had issued an export control directive requiring access to Fable 5 and Mythos 5 to be suspended for foreign nationals, whether they were inside or outside the United States. Anthropic said that enforcing the directive led it to disable access globally.
Reports linked the US government’s concern to the possibility that the models could be diverted to military intelligence users in countries including China or Russia. Another concern involved a possible jailbreak that could push the models towards identifying software vulnerabilities.
Why the AI decision is disputed
Anthropic pushed back, saying the government had not initially supplied specific technical details. The company characterised the reported examples as a narrow jailbreak rather than evidence of a universal capability that made the models uniquely dangerous. The report also notes that other AI models can assist with coding, security audits and red-team work.
The underlying issue is dual use. A capable code model may help security teams find and fix vulnerabilities, but it may also help attackers understand or exploit them. The report does not establish that Mythos 5 or Fable 5 are harmless, nor that they are uniquely dangerous compared with other models. It describes an unresolved question about whether a global withdrawal was a proportionate response based on the information publicly available.
TechManPat’s conclusion
I think the Australian SMS change is a useful, practical warning system, although users still need to verify messages independently. On the AI issue, I accept that powerful models can increase the capabilities of both defenders and attackers, but based on the public information described here, I am not fully convinced that pulling these models globally was justified as though they were uniquely dangerous.
This knowledge-centre summary is based on the linked TechManPat video and reflects the information available when it was published. Check current pricing, availability and policies before acting.
