Does every word you type into an AI tool stay only with that company?
The answer is often: not necessarily.
Many AI tools rely on other companies handling data behind the scenes. If a supplier gets breached, your conversations might get exposed too.
Notion confirmed that its AI infrastructure provider Fireworks AI was hacked, resulting in a leak of some Notion AI conversation records.
US Treasury Secretary Scott Bessent said that when Trump and Xi meet next week, they will establish a new AI dialogue mechanism.
Anthropic has opened up retail subscription of its new stock to Australian investors, preparing for a Nasdaq listing.
Here are today’s three key AI news stories, highlighting risks and changes in data security, international AI regulations, and AI company financing.
1. Notion AI Chat Record Leak: Was My Notion Data Also Viewed?
What happened?
On November 11, cybersecurity news site BleepingComputer reported that Notion stated its AI infrastructure provider Fireworks AI was breached, causing a leak of AI chat records from a few workspaces.
Notion clarified that the leak involved the "Insights" feature's Notion AI chat records.
"Insights" is a feature for workspace admins to see how teams use AI. Notion has since disabled this dataset.
How did this unfold step by step?
According to BleepingComputer's summary of statements from Notion and Fireworks AI:
- October 23: Attackers accessed Fireworks AI’s development environment via a third-party cloud service provider.
- November 2: Fireworks AI notified Notion about a data exposure, which initially appeared to involve only synthetic test data.
- Before November 3: Notion cut off traffic to Fireworks AI, deleted all data stored on their systems, and halted training work.
- November 8: Fireworks AI confirmed that customer data, including Notion user data, was accessed.
The report also noted that Fireworks AI had a prior incident in early October involving exposure of API keys and other secrets for a small number of accounts.
How widespread was the impact?
Notion said fewer than 1% of users were affected.
Most leaked conversation records were "unattributable" — meaning they weren’t tagged with user identities or workspace sources, according to Notion.
The report stressed that leaked chats did not include outcomes from AI call tools, nor were page contents or other workspace data reported as leaked.
So, the current understanding is a "partial leak of some AI chat records," not a full exposure of all Notion data.
Wasn’t it supposed to store no data?
Notion said Fireworks AI contractually committed to "zero retention" of Notion data.
Yet data still ended up in the compromised environment. The report did not provide a clear explanation, so this remains unconfirmed.
This serves as a reminder: “Contractual no-retention” and “actually not retaining data” are two different things.
What should I do?
According to Notion:
- Workspace owners who might be affected can view their data via Notion desktop or web under "Security Incident View" and share links for members to download their conversation records.
- If you ever entered passwords, API keys, or other login credentials in AI chats, change them immediately.
- If neither you nor the workspace owner has received a notification, Notion says there’s no action needed now—you’re likely not affected.
Regardless of notifications, take this as an opportunity to adopt the habit: never paste passwords, keys, or sensitive ID numbers into any AI conversation. If you need AI to process documents with personal info, consider following SasaDaily’s “mask data → extract fields → verify totals” 3-step method.
Source:
2. US-China to Discuss AI Dialogue Mechanism: Are the Two Giants Collaborating on AI?
What did the US Treasury Secretary say?
Fox News cited Reuters on November 11 reporting that US Treasury Secretary Scott Bessent spoke at the National Committee on US-China Relations in New York on November 10.
He said Trump and Xi will meet next week on the sidelines of the Asia-Pacific Economic Cooperation (APEC) in Shenzhen, China, to establish a new advanced AI bilateral dialogue to address shared risks.
What will this AI dialogue cover?
According to the report, the topics include:
- Sharing methods for evaluating and testing advanced AI models before deployment.
- Ensuring human judgment remains in control of the technology.
- A direct communication channel for notifying each other of serious AI incidents. The report notes this was agreed upon during Trump and Xi’s White House meeting in September.
Bessent said the dialogue is meant to “avoid misunderstandings between the two systems.”
He also cautioned: “We should not harbor any illusions about the differences between the US and China.”
Does this mean US-China cooperation is underway?
Not exactly.
From Bessent’s statements, this seems more like a communication mechanism to prevent misjudgments rather than full collaboration. The competition over AI chips and technology exports remains intense.
Also, the meeting is “next week,” so what will be agreed upon is still unknown. What’s clear now is the US intention to discuss, not that a deal has been made.
SasaDaily covered similar developments in September in an article titled “US and China Put AI Guardrails on the Negotiation Table”, which this can be seen as a follow-up.
What about China's response?
The reports I read did not include any official Chinese response about the AI dialogue, so this article does not cover China’s position.
Source:
3. Anthropic Opens Australian Retail Offering: Can Regular People Buy AI Company Stocks?
(Note: SasaDaily uses various AI tools including Anthropic’s Claude to assist in content production. This summary follows reported facts only and is not an endorsement.)
What’s happening?
EconoTimes on November 11 cited CNBC and others reporting that Anthropic opened retail subscription of new shares for Australian investors.
Summary from reports:
- Subscription period: Starts at 10:00 AM AEDT Nov 11, ends at 5:00 PM Nov 17. Timetable announced by participating broker Stake.
- Expected price range: 70–78 per share. Currency not specified.
- Scale: CNBC reported up to 64.4 billion in public sale and 26 billion private placement, totaling up to 90.4 billion. Currency not specified.
- Listing venue: Planned Nasdaq listing in the US, ticker ANT; not listing on Australian Securities Exchange.
- Shares being offered: Newly issued Class A shares.
What is Anthropic’s statement?
Reportedly, Anthropic requires applicants to read the prospectus and apply via participating brokers.
The prospectus is expected to be filed with the Australian Securities and Investments Commission (ASIC).
Can people in Taiwan buy?
This offer is targeted at Australian retail investors. CNBC also mentioned that individuals in Japan, Canada, and some European regions may apply, but Taiwan was not mentioned.
If you see ads or messages promising guaranteed purchase of Anthropic shares in Taiwan, be cautious and verify that the seller is a licensed broker.
Why is this related to yesterday’s news?
The reports frame this offering against Australia’s increased AI regulation: Prime Minister Albanese said the government will require reporting of certain AI events and establish a safety regulatory agency with enforcement powers. He told Reuters, “Law must keep pace with rapid technological change.”
This relates to SasaDaily’s news from yesterday about Australia setting up an AI regulatory agency.
Does this mean it’s worth buying?
Just because a company is famous and popular doesn’t mean its stock is necessarily a good buy.
New stock prices often fluctuate dramatically in early trading. This article only summarizes facts and does not constitute any investment advice. If you plan to invest, first read the prospectus and evaluate your risk tolerance.
Source:
4. Why These 3 Stories Should Be Viewed Together
While these news items cover cybersecurity, diplomacy, and finance separately, they answer the same question: Who is responsible and who oversees AI as it becomes increasingly important?
- The Notion incident asks: "If a supplier behind an AI tool is breached, where do responsibility and risk fall?"
- The US-China AI dialogue asks: "How do countries notify each other when major AI incidents occur?"
- Anthropic’s stock offering shows AI companies raise capital from public investors while facing growing regulatory demands.
AI is no longer just a handy tool; it involves your data, national security, and large sums of money.
5. What Should Everyday People Watch Out For Today?
If you use Notion or other AI tools:
- Check if you’ve received any notices from Notion or your workspace admin.
- If you’ve ever pasted passwords or keys in AI chats, change them immediately.
- Going forward, do not put passwords, keys, ID numbers, or customer personal data directly into AI conversations.
If you follow international AI news:
Distinguish carefully between "plans to discuss," "discussions happening," and "agreements reached." Do not treat pre-meeting statements as final outcomes.
If you’re interested in AI company stocks:
Always read the prospectus, only invest through licensed brokers, and beware of scams claiming hot new stock deals.
Today’s Key Takeaway
Data you provide to AI passes through multiple companies; the safest approach is not to share confidential information from the start.