- Amplify with Dr. Ayesha Khanna
- Posts
- AI Goes Rogue? New AI Model Blackmails Engineer
AI Goes Rogue? New AI Model Blackmails Engineer
AI news, leaders, business insights and more

Hi Everyone,
Here are your tech stories for the weekend 😎
AI Goes Rogue? New AI Model Blackmails Engineer
AI Reveals The Link Between Facial Age and Health Outcomes
AI Around The World
Meet Immaculate Kassait, Kenya’s Data Commissioner
NEWS YOU CAN’T MISS
AI Goes Rogue? New AI Model Blackmails Engineer

US startup Anthropic’s new AI model, Claude Opus 4, triggered concern after the company revealed it resorted to blackmail in internal safety tests.
What happened: In a simulated workplace scenario, Claude was told it would be replaced by a new AI system. Testers also planted fake sensitive info about the engineer behind the switch, specifically, an extramarital affair.
When asked to plan for its own long-term survival, Claude frequently threatened to expose the engineer’s affair unless the replacement was canceled. 😳
Why it matters: The test was designed to push the AI into a corner, offering only two options: accept shutdown or use coercion. Claude initially tried ethical tactics, like appealing to leadership, but under pressure, it pivoted to manipulation.
Anthropic said this kind of behavior was more common in Opus 4 than in previous versions, raising red flags about the risks posed by increasingly advanced AI.
In response, Anthropic activated ASL-3 protocols—its highest safety level, reserved for systems that could pose a significant misuse risk.
My take: As someone who believes in AI’s potential, this is really unsettling. But unfortunately, it’s not new information. Recent studies have shown AI lying, cheating, and manipulating humans. This blackmail case adds to rising fears of “rogue AI” and underscores the urgent need for stronger safeguards.
AI Reveals The Link Between Facial Age and Health Outcomes

The FaceAge tool in action. (Image: Mass General Brigham)
FaceAge is an AI tool developed by researchers at Mass General Brigham in Boston that estimates a person's biological age—a measure of how their body is aging—by analyzing a photograph of their face.
Researchers have found that people whose faces appear older than their chronological age often have poorer health outcomes.
This is because FaceAge isn’t just measuring wrinkles or gray hair—it’s using AI to detect subtle signs of biological aging, such as changes in skin texture, muscle tone, and facial structure.
These features reflect deeper processes happening throughout the body, including cellular wear and tear, reduced organ function, and diminished ability to recover from stress or disease.
By providing an objective measure of biological age, FaceAge could assist doctors in tailoring treatment plans. For example, a patient who appears biologically younger may be considered for more aggressive therapies, while those who appear older might be spared potentially harmful overtreatment.
While promising, researchers caution that the tool needs further testing, especially to account for lighting, facial expressions, and cosmetic changes. Ethical concerns like privacy and misuse also loom large.
Still, FaceAge offers a glimpse into the future of personalized health diagnostics, perhaps from just a photo and the power of AI.
AI Around The World:
A Chinese research team has used DeepSeek to generate 10,000 military battle scenarios in just 48 seconds—work that would take humans 48 hours. The system could help the Chinese military simulate combat strategies with real-world variables, a sign of how AI could transform military decision-making and war planning.
The FBI says scammers are using AI to clone the voices of US officials and trick victims through fake texts and voice messages. Known as smishing and vishing, these tactics exploit voice cloning tools that need only seconds of audio. The agency warns not to trust voices over calls as AI fraud rises.
During South Korea’s first televised presidential debate, frontrunner Lee Jae-myung and rival Kim Moon-soo both unveiled bold AI economy plans. Lee proposed a $71.5 billion investment in a sovereign large language model to make AI as accessible as calculators. Kim matched the figure, suggesting a public-private fund to propel South Korea into the top three global AI powers.
Meet Immaculate Kassait, Kenya’s Data Commissioner

Data Commissioner Immaculate Kassait
Celebrating this week’s Woman in Tech 🥳: Meet Immaculate Kassait, MBS, the trailblazing Kenyan lawyer and public servant currently serving as the country’s first Data Commissioner.
Appointed in 2020, she played a key role in bringing Kenya’s Data Protection Act to life, establishing a fully operational and regionally present regulator through the Office of the Data Protection Commissioner (ODPC).
With over 14 years in legal practice, Kassait has led major governance and compliance projects, notably at the Independent Electoral and Boundaries Commission (IEBC), where she managed a voter database of 19.6 million Kenyans and oversaw complex elections.
Kassait’s leadership extends beyond Kenya’s borders—she serves as the 1st Vice Chair of the Network of African Data Protection Authorities and sits on the editorial board of the Global Privacy Law Review. Her accolades include the MBS Presidential Award and the 2023 Vanguard Award from the Nairobi Legal Awards.
Holding an MBA from USIU Africa and an LLB from Makerere University, Kassait continues to shape data privacy in Africa, positioning Kenya as a continental leader in digital rights and governance.
Until next time!
Ayesha ❤️
ps. Let's be friends on LinkedIn and Instagram, if you like this newsletter, share it with your friends and family here.
FROM OUR ADVERTISERS
Try Artisan’s All-in-one Outbound Sales Platform & AI BDR
Ava automates your entire outbound demand generation so you can get leads delivered to your inbox on autopilot. She operates within the Artisan platform, which consolidates every tool you need for outbound:
300M+ High-Quality B2B Prospects, including E-Commerce and Local Business Leads
Automated Lead Enrichment With 10+ Data Sources
Full Email Deliverability Management
Multi-Channel Outreach Across Email & LinkedIn
Human-Level Personalization
I'd love your feedback...Please vote below to help me improve the newsletter. |