Friday’s AI Report
• 1. 💡 Microsoft reducing reliance on OpenAI?
• 2. 💰 Learn about the cost of data breaches with IBM
• 3. 🌍 How Ford used AI to cut costly vehicle re-calls
• 4. 🤖 Build advanced AI chatbots with ChatNode
• 5. 💡 Partner Perspectives
• 6. ⚙️ Trending AI tools
• 7. 💾 Anthropic to store chats for 5 years!
• 8. 🌮 Taco Bell halts AI drive-thru
• 9. 📑 Recommended resources
Read Time: 5 minutes
✅ Refer your friends and unlock rewards. Scroll to the bottom to find out more!
Microsoft reducing reliance on OpenAI?
🚨 Our Report
Microsoft has launched its first in-house AI models, MAI-Voice-1 (a speech generation model) and MAI-1-preview (a text-based model for Copilot), both of which are primarily focused on efficiency and consumer use cases.
🔓 Key Points
MAI-Voice-1, which is already powering Microsoft’s Copilot Daily (audio news summaries) and AI podcast features, can generate 1 minute of audio in under 1 second and allows for voice and speaking style customization.
MAI-1-preview is currently powered by OpenAI technology and provides “helpful [text] responses to everyday [text] queries” and, following user testing, will be integrated into Copilot to support selected Copilot tasks.
The new models have been developed around efficiency/cost-effectiveness: MAI-Voice-1 runs on just 1 chip, MAI-preview-1 was trained on 15k chips—significantly lower than rivals (eg. Grok was trained on 100k+).
🔐 Relevance
Although MAI-preview-1 still relies on OpenAI tech, by releasing these models, it’s clear that Microsoft is planning to reduce its reliance on OpenAI and establish a greater level of independence within the AI space. Its AI Chief, Mustafa Suleyman, confirmed that their focus was on consumer use, not Enterprise (a market OpenAI is chasing), as they “have vast amounts of very predictive and very useful data” on ads and how consumers use their products/services. He also believes that, at a time when concerns are growing over an AI bubble, brought on by escalating development and training costs, efficiency and cost-effectiveness are crucial to remain sustainable.
New report reveals insights on AI-driven attacks
Our study found AI-driven attacks accounted for 1 in 6 data breaches.
Attackers can use generative AI to polish and scale their phishing campaigns and other social engineering attacks.
Gen AI can reduce the time needed to craft a convincing phishing email from 16 hours down to just five minutes on average.
10 Actions You Can Take to Prevent AI Litigation
By Karen L. Odash · Labor And Employment Attorney at Fisher Phillips
The AI Report Partner Perspectives Column
“The issue isn’t just bad data—it’s when algorithms quietly turn old biases into new barriers.”
A job applicant is suing a company he interviewed with, because he felt their AI-powered hiring system discriminated against him by embedding racial bias from historical data. The lawsuit highlights growing risks for employers relying on automated hiring tools. This is becoming a broader trend in workplace AI litigation.
How Ford used AI to cut costly vehicle re-calls and re-work
Ford faced escalating re-call and re-work costs due to hard-to-detect assembly errors eg. a fuel leak recall affected 600,000 SUVs and cost $570M
They deployed an in-house AI Computer Vision System, across its assembly lines, that detected millimeter-scale misalignments in real-time.
They also built an AI Mobile Assembly Inspection Vision System that used smartphone cameras to verify proper part installation during assembly.
As a result, they could catch assembly issues before they escalated into costly recalls or rework and minimized the need for invasive, expensive fixes.
This AI Solves Support Tickets Before They Open
Your leads vanish when they’re left on hold or met with slow replies.
Add Chatnode to your website for:
24/7 human-like replies that feel natural
Automatic knowledge updates to ensure your AI delivers its best
Zero coding, so you can launch custom AI that understands your business
Live analytics that reveal user behavior and conversation trends
Unlike typical enterprise solutions that demand four/five-figure budgets, ChatNode is affordable and offers the latest in AI-driven customer support.
Your first month is on us ➝ use code: AIREPORT
Mindstream helps readers stay on top of the rapid developments in AI and integrate it into their work. Join over 200,000 subscribers ⭐️⭐️⭐️⭐️⭐️
Mobser delivers intelligent AI-powered data analytics
Stakly builds production-ready web apps with AI
Anthropic will start training its Claude AI models on user data—including new/resumed chats and coding sessions (older chats that haven't been revisited will not be affected)—by default, unless users opt out by 28th Sep.
Those who don’t choose to opt out will also have their data (from new/resumed chats/coding sessions) stored for five years; Anthropic previously deleted user data after 30 days.
Anthropic is making these changes to “improve model safety,” making their “systems for detecting harmful content more accurate and less likely to flag harmless conversations,” and also to improve model skills.
After taking 200M food orders via Alexa-like AI voice assistants from its drive-thru windows, Taco Bell has decided to re-evaluate its ambitious AI plans after “learning a lot” and feeling “let down” by AI’s capabilities.
According to several frustrated customers, the AI drive-thru assistants would often get their orders wrong or get caught in a loop, asking the same questions—eg. ”What would you like to drink?”—over and over again.
Some were asking the AI for things like “18,000 cups of water”, or a “McDonald’s Big Mac,” which it happily accepted, forcing a human drive-thru assistant to step in—begging the question: Why is AI needed?
MORE NEWS
PODCASTS
Why you should ignore MIT’s viral AI “study”
This podcast dissects MIT’s recent study—which states that 95% of Enterprise AI pilots fail—which has created havoc within the AI industry, with people fearing we’re in an AI bubble that’s about to burst.
We read your emails, comments, and poll replies daily.
Until next time, Martin, Liam, and Amanda.
P.S. Unsubscribe if you don’t want us in your inbox anymore.