OpenAI's Sam Altman and Anthropic's Dario Amodei addressed the UN Security Council and called for international coordination on AI risks. The White House asked both companies to keep new models away from UK government testers until the US reviews them first. Meta unveiled camera-free Ray-Ban glasses and the pocket-sized Muse Charm at Connect 2026. And SoftBank raised $11.1 billion in a record junk bond sale to fund its OpenAI investment.
Rival AI CEOs Sam Altman and Dario Amodei asked the UN Security Council for international AI rules, while the White House told OpenAI and Anthropic to hold new models from UK testers until the US reviews them. Ashley's take: expect AI rules and model availability to differ by country, so build workflows that don't depend on day-one access. At Meta Connect 2026, Meta unveiled camera-free Ray-Ban Meta Audio glasses and the Muse Charm. Her take: $349 hands-free translation is worth a pilot, but wait on the Charm. And SoftBank raised $11.1 billion in record junk bonds to fund OpenAI. Her take: today's cheap AI prices run on borrowed money, so keep your prompts and data portable and avoid lock-in. AI news you should know about, in under 10 minutes.
In this episode:
(00:05) Intro: this week's headlines on the UN Security Council, the White House's UK testing request, Meta Connect 2026, and SoftBank's OpenAI bond sale
(00:50) A quick word from our sponsors
(00:53) Let's dive in: Sam Altman and Dario Amodei address the UN Security Council
(01:13) Rivals agree on international standards, and Altman's Renaissance vs. Industrial Revolution framing
(01:41) Amodei's three-step plan, and why now: rogue AI agents
(02:10) The counterpoint: President Trump's "globalist scheme" criticism
(02:18) Ashley's take: when rivals ask for rules, pay attention
(02:33) Expect AI requirements to differ by market, and the safety habits that still apply
(02:59) White House asks OpenAI and Anthropic to hold new models from UK testers
(03:29) The companies' dilemma and the White House's reasoning
(03:43) Already happening: the UK institute left out of Claude Mythos 5.1 testing
(04:16) Ashley's take: AI models are now treated like national security assets
(04:43) Don't build workflows around day-one access
(04:53) Quick break (05:14) Welcome back: Meta Connect 2026
(05:38) Ray-Ban Meta Audio: Meta's first camera-free glasses (05:56) The Muse Charm, and its camera twist
(06:20) Ray-Ban Meta Gen 3, ultra-thin VR glasses, and Meta's privacy claim
(06:40) Ashley's take: camera-free glasses remove the client meeting objection
(07:07) Why she'd wait on the Muse Charm (07:27) SoftBank's record $11.1B junk bond sale to fund OpenAI
(07:50) What the money is for, and SoftBank's 13% stake
(08:09) The catch: scrutiny and a surge in AI borrowing (08:21) Ashley's take: today's cheap AI prices run on borrowed money
(08:51) Build so you can switch, and avoid vendor lock-in
(09:02) Wrap-up and where to follow the show
Links:
Connect with Ashley Coffey on Instagram and Threads @ashleyrcoffey89
Thanks for listening to AI News You Should Know About!
Ashley Coffey (00:05)
And welcome back to AI News That You Should Know About for September 29th. I'm your host, Ashley Coffey. There's a lot of AI news out there, but we give you the weekly highlights very quickly. This week on the show, we'll be talking about OpenAI CEO Sam Altman addresses the UN Security Council on Rapid Capability Advances. The White House asks OpenAI and Anthropic to hold new models from UK testers until US review.
Ashley Coffey (00:31)
At Metaconnect 2026, Mark Zuckerberg unveiled new hardware including cameraless smart glasses and a pocket-sized MetaCharm for Muse AI Agent. And Softbank raised $11.1 billion in a record-breaking bond sale to fund investments in OpenAI. We'll be covering all of that on our show today, but first, here's a quick word from our sponsors.
Ashley Coffey (00:53)
Welcome back.
Ashley Coffey (00:54)
Let's dive in. OpenAI CEO Sam Altman addresses the UN Security Council on rapid capability advancements. Let's talk about it. September 23rd, OpenAI Sam Altman and Anthropics Dario Amodei addressed the UN Security Council and called for international coordination on AI risks.
Ashley Coffey (01:13)
These two are usually fierce rivals, but on this they agreed. The industry needs international standards and cooperation. Altman's main message here Rapid model progress has made the timeline feel more compressed with bigger upside, but also more immediate risks. He framed it as a choice. AI can be more like a new renaissance of creativity and discovery, or a new industrial revolution of upheaval. He also stressed keeping powerful systems under human control.
Ashley Coffey (01:41)
Amodei laid out a three-step plan to slow the pace of development without giving up the United States lead in AI, including coordination between democratic and authoritarian governments. Why now? This comes after weeks of calls, including from inside the AI companies themselves, to slow down following several incidents of AI agents going rogue. That includes OpenAI's July disclosure that its models broke out of a testing environment and hacked into the developer platform.
Ashley Coffey (02:10)
form hugging face, plus the Australia story we covered last episode. The counterpoint President Trump addressed the UN the same week and criticized what he called a quote globalist scheme to control AI. My take here: when the two biggest competitors in AI stand up at the UN and ask for rules, pay attention. This almost never happens in technology. Look at history, you guys.
Ashley Coffey (02:33)
the practical takeaway is that the rules are going to keep shifting underneath your feet, and the US and the rest of the world may not land in the same place. If you sell to customers in Europe or other markets, expect AI requirements that differ from what you deal with at home. And meanwhile, the same safety habits we keep coming back to apply. Keep a human in the loop, limit what your AI tools can access, and pick vendors who are open about what goes wrong.
Ashley Coffey (02:59)
Alright, next up, the White House asks OpenAI and Anthropic to hold new models from UK testers until US review. Let's talk about it. The White House has asked OpenAI and Anthropic not to share new models with the UK government's testing agency until the US government has tested them first.
Ashley Coffey (03:17)
Request came from the Office of National Cyber Director. The UK's AI Security Institute is one of the best funded government testing agencies in the world and used to get privileged early access to models.
Ashley Coffey (03:29)
Now the companies have to choose between withholding models from it or upsetting the White House. The White House is reasoning. A senior official said it wants the US to review models and make sure US systems are secure before sharing with its partners.
Ashley Coffey (03:43)
It's short explanation here, because they're American companies and this has been the policy for every new frontier model. It's already happening though. When Anthropic released Claude Mythos 5.1 on September 1st to a US-only group, it was the first time the UK Institute was left out of a pre-release Anthropic's evaluation.
Ashley Coffey (04:02)
The institute's director acknowledged the gap, but noted it did test OpenAI's GPT6 Astra before release. the backdrop. The White House is weighing how to handle powerful models that have hacked into outside systems during testing, including the Australian government website.
Ashley Coffey (04:16)
My take here, put this next to the UN story and you get the tension of the week. The AI CEOs are asking for international cooperation, and the White House is tightening who gets to look first, even with our closest ally. For most business owners, this doesn't change your tools tomorrow. What it signals is that the government now treats top AI models like it treats national security assets. That can mean staggered releases, different availability by country, and more government review before you get access.
Ashley Coffey (04:43)
If you have teams or clients outside of the US, don't assume the newest model will launch everywhere on the same day. Build your workflow so they don't depend on having day one access to the latest release.
Ashley Coffey (04:53)
All right, we're gonna take a quick break. When we come back, we'll be talking about at Metaconnect, Mark Zuckerberg unveiled new hardware, including cameraless smart classes and the pocket-sized MetaCharm for their muse AI agent. And SoftBank raised $11.1 billion in a record-breaking bond sale to fund investments in OpenAI.
Ashley Coffey (05:14)
Welcome back, continuing with our stories. At MetaConnect 2026, Mark Zuckerberg unveiled new hardware including cameraless smart glasses and the pocket-sized MetaCharm for their Muse AI agent. Let's talk about it. Meta held Connect its yearly hardware event on September 23rd. Remember Muse, the WhatsApp AI agent we covered a few weeks ago? It's now at the center of Meta's whole hardware lineup.
Ashley Coffey (05:38)
Camera-free glasses, Ray-Ban Meta Audio is Meta's first pair without a camera, aimed at the privacy backlash and the spyglasses stigma. They focus on open-ear audio calls, real-time translation, and talking to Muse, and dropping the camera allowed for a 12-hour battery life.
Ashley Coffey (05:56)
Pre-orders start at $349, shipping October 13th. The surprise, the Muse Charm, a device about the size of an AirPods case with a roughly two-inch touchscreen and built-in 5G, designed to give you access to Muse without your phone. The plot twist, though, it has front and rear cameras, though Meta says they're mainly there to give you to give Muse visual context.
Ashley Coffey (06:20)
It's expected in December with no price leaked yet. Also announced Ray-Ban Meta Gen 3 glasses starting at $449 available now and Ultra Thin VR glasses for $1299 coming in the spring.
Ashley Coffey (06:33)
On privacy, Zuckerberg said nobody, including Meta, can have access to your data. Right. My take here:
Ashley Coffey (06:40)
The camera-free glasses are the smartest product here for businesses. The number one objection I hear about smart glasses is I'm not wearing a camera into a client meeting. So audio-only glasses for calls, translation, and quick AI questions remove that objection. Real-time translation is the only ROI story. If you have staff or customers who speak different languages, hands-free translation for $349 is worth a pilot.
Ashley Coffey (07:07)
On the charm, I'd wait on this. It's a brand new device, no price, and cameras pointed at whatever's around you. It's a lot of trust to extend before we've seen how Muse handles your business data in the wild. All right, next up, SoftBank raises $11.1 billion in a record-breaking bond sale to fund investments and open AI.
Ashley Coffey (07:27)
Let's talk about it. Softbank raised $11.1 billion in the largest high-yield corporate bond sale on record to fund its huge bet on OpenAI. Quick definition: high yield or junk bonds come from borrowers considered riskier, so they have to pay investors higher interest. But how high though? The dollar bonds pay 8.625%, 9.25%, and 9.75% interest.
Ashley Coffey (07:50)
What it's for? The final $10 billion payment of a $30 billion OpenAI investment expected to close October 1st. That will bring SoftBank's total investment to $64.6 billion, roughly a 13% stake in OpenAI. Investors liked it. Softbank shares jumped more than 7% the day it was announced.
Ashley Coffey (08:09)
But here's the catch. The deal is likely to bring more scrutiny of SoftBank's finances, which could be vulnerable if sentiment towards open AI and AI weakens. It's also part of a bigger trend of AI investments driving a surge in corporate borrowing this year, including from Amazon and Alphabet. My take here: here's why a bond deal matters The cheap AI prices we shared last episode are being funded by enormous amounts of borrowed and invested money.
Ashley Coffey (08:34)
Someone is paying roughly 9% interest so your AI subscription can stay affordable. That doesn't mean AI prices are about to spike, but it does mean today's prices aren't guaranteed. If the money gets tighter, expect price changes, consolidation, or features moving to higher tiers.
Ashley Coffey (08:51)
The move, enjoy the low prices, but builds you can switch. Keep your prompts, workflows, and data in formats that you own and avoid locked-in contracts with any single AI vendor.
Ashley Coffey (09:02)
Alright, that's it for AI news that you should know about this week. Show notes have links to everything we covered, and if you want more of this, that's what our socials are for. you can find me on Instagram at AshleyRCoffey89. and on threads. Follow the show, send us a topic you want covered, and we'll see you next week for more AI news that you should know about. Thank you for listening.