Why Cohere is merging with Aleph Alpha
Canadian AI startup Cohere is taking over Germany-based Aleph Alpha, with the blessing of their governments, in a bid to offer a sovereign alternative to enterprises in an AI landscape dominated by American players. “Sovereign AI” refers to systems where companies and governments retain full control over their own data — rather than routing it through U.S. tech giants like Microsoft or Google. ...
While Cohere reported $240 million in annual recurring revenue in 2025, Aleph Alpha had previously generated little revenue and significant losses. But investors are betting that teaming up will improve their odds against much larger rivals. ...
The new entity plans to target highly-regulated industries — including defense, energy, finance, healthcare, manufacturing and telecommunications— as well as the public sector. ...
According to Gomez, “Cohere will become a Canadian-German company.” But that promise could be harder to keep if the company goes public — putting ownership in the hands of global shareholders with no particular allegiance to either country.
See the full story here; https://techcrunch.com/2026/04/25/why-cohere-is-merging-with-aleph-alpha/
The core thesis: this isn’t just a merger—it’s a geopolitical AI move
The merger between Cohere and Aleph Alpha is fundamentally about “sovereign AI”—not just scaling models.
- Governments (especially in Europe) want AI systems they control
- Enterprises in regulated sectors want data residency + compliance guarantees
- There’s growing resistance to dependence on U.S. hyperscalers
...
See the full story here: https://chatgpt.com/?utm_source=Daily+Email+2025&utm_campaign=dd635137fe-EMAIL_CAMPAIGN_2026_04_27_01_24&utm_medium=email&utm_term=0_-dd635137fe-303719286&mc_cid=dd635137fe&mc_eid=116e9f337b&model=gpt-5-3
How Anthropic Thinks About Agents, Workflows, and Tasks
...
Task – A single model call: Summarize this. Classify that. Extract these fields. Two years ago this felt like magic. Today it is table stakes. The cost is predictable and the failure modes are bounded.
Workflow – Multiple model calls in a predefined control flow. You decide the steps, the model fills them in. Route the customer email to the right queue, draft a response, check the response against policy, then send. The control flow is yours. The intelligence happens at each node.
Agent – A model using tools in a loop, deciding its own trajectory. You give it a goal, a set of tools, and a system prompt. It decides what to do next based on what just happened. The control flow belongs to the model. ...
China blocks Meta from acquiring AI startup Manus
...
In a one-line statement, China's National Development and Reform Commission, the country's top planning agency, said it was prohibiting a foreign acquisition of Manus and had required all the parties to withdraw from the deal. It did not specifically name Meta Platforms, which owns Facebook and Instagram.
The decision was made by the commission's Office of the Working Mechanism for Security Review of Foreign Investment in accordance with Chinese laws and regulations, the statement said. ...
Meta announced in December that it was acquiring Manus, which has Chinese roots but is based in Singapore, in a rare case of a major U.S. tech group buying an AI company with strong links to China. Its deal with Manus, whose "general-purpose" AI agent can perform multistep complex work autonomously, was expected to help expand AI offerings across Meta's platforms. ...
See the full story here: https://www.npr.org/2026/04/27/g-s1-118892/china-blocks-meta-from-acquiring-ai-startup-manus
The Sphere Looked Like a Disaster. It’s Become a Huge Hit Instead.
...
Sphere has discovered an unexpected hit formula: the performance space that is the most cutting-edge technically works with bands whose frontmen are old enough to have decades of greatest hits that everyone knows by heart. And it can fill the arena with fans the age of its stars—U2’s Bono (65)—as well as vintage-curious Gen Z.
It stacks the calendar with established artists for residencies that can last days, weeks or months—pairing safe, broadly appealing musical acts with spectacular visual flair in shows that take months to produce: U2, Eagles, Kenny Chesney and Backstreet Boys, among others. During the day before a concert, it can charge $200 a ticket for “The Wizard of Oz,” remastered so audiences feel like they’re walking with Dorothy on the yellow brick road. ...
See the full story here: https://www.wsj.com/business/media/sphere-vegas-dolan-disaster-hit-fa0e6b17
AI remains lacking in clinical reasoning abilities, according to study of 21 large language models
Despite increasing use of artificial intelligence (AI) in health care, a new study led by Mass General Brigham researchers from the MESH Incubator shows that generative AI models continue to fall short in their clinical reasoning capabilities. ...
By asking 21 different large language models (LLMs) to play doctor in a series of clinical scenarios, the researchers showed that LLMs often fail at navigating diagnostic workups and coming up with a testable list of potential or "differential" diagnoses.
Though all tested LLMs arrived at a correct final diagnosis more than 90% of the time when provided with all pertinent information in a patient's case, they consistently performed poorly at the earlier, reasoning-driven steps of the diagnostic process, according to results published in JAMA Network Open.
"Despite continued improvements, off-the-shelf large language models are not ready for unsupervised clinical-grade deployment," said corresponding author Marc Succi, MD, executive director of the MESH Incubator at Mass General Brigham. ...
See the full story here: https://medicalxpress.com/news/2026-04-ai-lacking-clinical-abilities-large.html
The AI Brain That Gets Smarter by Shrinking
Summary: In the world of AI, bigger is usually seen as better—but this leads to massive energy consumption and computational costs. Taking a cue from human biology, a research team has developed a brain-inspired “selective pruning” framework for Spiking Neural Networks (SNNs).
The study reveals that AI doesn’t need more connections to learn complex tasks; it needs the right ones. By mimicking how an infant’s brain strengthens long-range links while “pruning” away local clutter, this new AI achieves continual learning—mastering perception, motor control, and interaction—while actually getting smaller and more energy-efficient over time.
Key Facts
- The “Infant” Approach: Human brains don’t just add connections; they refine them. This model follows a “simple-to-complex” trajectory, maturing primary modules (like perception) before moving on to higher cognition.
- Selective Pruning: Unlike traditional AI that freezes weights to prevent forgetting, this system introduces a feedback mechanism that actively inhibits and removes redundant local connections from earlier tasks.
- Knowledge Reuse: While local clutter is pruned, cross-regional “long-range” connections are strengthened. This allows the AI to reuse knowledge from old tasks to solve new ones without needing more “brain” space.
- No More “Catastrophic Forgetting”: A major hurdle in AI is that learning something new often “erases” the old. This developmental framework mitigates that loss without using energy-heavy tricks like “experience replay.”
- Sustainably Evolving: The network scale is continuously reduced as learning progresses, offering a low-energy pathway toward General Cognitive Intelligence.
See the full story here: https://neurosciencenews.com/brain-inspired-pruning-continual-learning-30497/
Elon Musk’s xAI sues Colorado over state’s new AI law
...
The law imposes disclosure and risk-mitigation requirements on developers of so‑called "high‑risk" AI systems used in decisions involving employment, housing, education, health care and financial services. ...
Elon Musk's artificial intelligence firm said the law violates the First Amendment by restricting how developers design AI systems and compelling speech on contentious public issues.
The company says the law would force it to alter its flagship AI model, Grok, to reflect the state's views on diversity and discrimination rather than being objective. ...
While some tech companies and Republican lawmakers want states to leave AI regulation to Washington, California's attorney general has warned against relying solely on Congress, pointing to years of delays on data privacy and technology laws. ...
See the full story here: https://www.reuters.com/legal/government/elon-musks-xai-sues-colorado-over-states-new-ai-law-2026-04-09/
Major Federal AI Developments Signal a Rapidly Shifting Regulatory Landscape
...
I. The Trump America AI Act – Sweeping Federal AI and Platform Reform Proposal
Blackburn has released a discussion draft of The Republic Unifying Meritocratic Performance Advancing Machine Intelligence by Eliminating Regulatory Interstate Chaos Across American Industry Act (the “Trump America AI Act”). The bill is expansive in scope, combining AI-specific requirements with broader reforms to Internet liability, platform governance and digital infrastructure.
Key Elements of the Bill
Repeal of Section 230
- The bill would repeal Section 230 of the Communications Act of 1934, (18 USC Sec. 230) after a two-year transition period, reversing the broad immunity that has been historically enjoyed by social media platforms and other communications entities over liability for content.
- This change would significantly expand potential liability for online platforms with potential downstream effects on moderation practices and platform design. Namely, companies may need to adjust moderation policies and methods or reduce reliance on engagement-driven ranking features.
...
Although the TRUMP AMERICA AI Act has not been formally introduced and is unlikely to advance in its current form, it provides a useful roadmap of policy directions that may shape future federal proposals and state-level legislation. ...
See the full story here: https://www.manatt.com/insights/newsletters/manatt-digital-and-technology/major-federal-ai-developments-signal-a-rapidly-shifting-regulatory-landscape
When Flock Comes to Town: Why Cities Are Axing the Controversial Surveillance Technology
Early this year, my home city of Bend, Oregon, ended its contract with surveillance company Flock Safety, following months of public pressure and concerns around weak data privacy protections. Flock's controversial AI-powered license plate cameras were shut down, and its partnership with local law enforcement ended. ...
Though Flock doesn't have a direct partnership with federal agencies (a blurry line I'll discuss more), law enforcement agencies are free to share data with departments like ICE, and they frequently do. ...
Following Super Bowl ads about finding lost dogs, Flock was under scrutiny about its planned partnership with Ring, Amazon's security brand. The integration would have allowed police to request the use of Ring-brand home security cameras for investigations. Following intense public backlash, Ring cut ties with Flock just like my city did. ...
See the full story here: https://www.cnet.com/home/security/when-flock-comes-to-town-why-cities-are-axing-the-controversial-surveillance-technology/
Whose AI Is It Anyway? Key Developments in the Evolving Relationship Between AI and IP
- Divergence across major jurisdictions on key issues such as the protectability of AI tools and outputs, the use of copyrighted data for AI training, and the requirements for inventorship and authorship require companies to adopt different IP strategies in different countries.
- In the U.K. and EU, the scope for patent protection of AI tools has broadened, which may be prompting some companies to shift from relying on trade secret and copyright protections toward implementing formal patent strategies.
- The use of copyrighted works as AI training data is a major area of contention.
- The proliferation of AI-generated voices and images has intensified concerns over digital identity rights. Celebrities and notable figures are exploring a patchwork of different protections.
See the full story here: https://www.skadden.com/insights/publications/2026/04/insights-april-2026/whose-ai-is-it-anyway
Pages
- About Philip Lelyveld
- Mark and Addie Lelyveld Biographies
- Presentations and articles
- Trustworthy AI – A Market-Driven approach
- Tufts Alumni Bio