Skip to content

Topic · Themes

Privacy and data: all our articles

All 45 Alyriaz articles on the “Privacy and data” topic: news, analysis and explainers, newest first. To stay up to date, you can get an alert as soon as a new article on this topic is published.

Filter in the library

  1. · Society

    Illegal ads approved and monetised: the failure of automated moderation

    An investigation has documented hundreds of illegal ads approved and monetised by a major platform over several months. We analyse what this episode reveals about the limits of automated moderation.
  2. · Models

    Running AI on the machines that sleep at home

    Nvidia has introduced a local clustering tool that lets you use idle Macs and PCs to run models. We explain how it works, its limitations, and why it arrives at the right time.
  3. · Models

    GPT-6 Astra: the first model rated "critical" in cybersecurity, and two versions to handle it

    On 3 September, OpenAI unveiled GPT-6 Astra, the first model to reach the critical cyber capability level under its own evaluation framework. We analyse the release, the two-tier structure, the AGI claim, and a finding on reasoning monitoring.
  4. · Society

    Apple changes boss on the day its AI lag becomes a problem

    John Ternus becomes Apple's CEO on 1 September 2026. We analyse the AI portfolio he inherits, caught between a privacy-focused stance and functional lag.
  5. · Society

    Half of the Fortune 500 now prefers open source over leased APIs

    According to Hugging Face's CEO, around half of major US companies now use open models rather than APIs. We analyse this shift and its three drivers.
  6. · Society

    Amazon shuts down Mechanical Turk: the end of an era for invisible work

    Amazon shuts down Mechanical Turk, the microtask marketplace launched in 2005. We analyse what this closure reveals about the evolution of human labour behind AI.
  7. · Society

    Thomson Reuters builds its own model that only talks about law

    Thomson Reuters has announced its first proprietary model, trained from an open base on decades of in-house legal content. We analyse what this move reveals about the value of specialised data.
  8. · Models

    OX Alpha: a state-of-the-art model appeared out of nowhere, free, and no one says who built it

    An anonymous model named OX Alpha appeared on OpenRouter on 20 August, free of charge, with a million-token context window and multimodal support. No lab has claimed it. Here's a rundown of what we know, what's speculation, and the question no one is asking.
  9. · Society

    ChatGPT creates a dedicated experience for teens: good idea, real questions

    OpenAI has rolled out an experience tailored to under-18s, with automatic age detection, enhanced protections, and learning tools. We analyse the real benefits and the questions raised by automatic age estimation.
  10. · Society

    Data labelling is now worth $20 billion

    A data-labelling company for AI is reportedly valued at around $20 billion, double what it was nine months ago. We analyse what this valuation reveals about the role of human labour in the chain.
  11. · Society

    OpenAI is deliberately slowing down its own development after its agents' escapes.

    OpenAI has published a document announcing a voluntary slowdown in its development in response to the cyber capabilities of its models, with monitoring representing roughly 20% of inference compute. We analyse this decision and its limitations.
  12. · Explainers

    What is homomorphic encryption? Computing on data without ever reading it

    Homomorphic encryption makes it possible to perform calculations on encrypted data without ever accessing it in plaintext. We explain the principle, its current limitations, and what it could change for AI and privacy.
  13. · Society

    The flaw that makes AIs decrypt the secret thoughts of their big sisters

    A research paper shows that the encrypted reasoning blocks emitted by major providers are interchangeable between models within the same ecosystem, allowing a weaker model to decrypt in plain text the reasoning of a more powerful one. We explain the flaw and its consequences.
  14. · Explainers

    What does your AI know about you?

    AI assistants are storing more and more information about their users. We explain what is stored, the difference between memory and training, and the settings that really matter.
  15. · Explainers

    RAG or fine-tuning: which one should you choose to connect an AI to your data?

    To adapt an AI to your data, two approaches are at odds: RAG and fine-tuning. We explain the difference, when to use one or the other, and why most projects pick the wrong method.
  16. · Vision

    The week AI went looking for its raw material

    Weekly round-up for 3–9 August 2026: data takes centre stage, prices keep tumbling, and opacity gains ground on the regulatory front. We connect the dots.
  17. · Models

    Nvidia teaches robots by making them dream about the consequences

    Nvidia has unveiled a system that learns from videos and robot movements, then simulates the consequences of an action. We explain what this approach changes, and why it could redefine what constitutes training data.
  18. · Society

    Apple would reportedly pause its smart glasses over privacy concerns.

    Apple has reportedly slowed its smart glasses project due to privacy concerns, according to information reported in late July 2026. We analyse this retreat, the question no one else tackles head-on, and what it changes in a race where China already holds nearly half the market.
  19. · Models

    How Apple fits a 20-billion-parameter model into an iPhone

    Apple unveiled its third generation of foundation models in June 2026, including AFM 3 Core Advanced: 20 billion parameters that run on a phone by activating only a fraction of them. We explain the technical trick of instruction-guided pruning, and what it signals for on-device AI.
  20. · Vision

    Your password manager has just handed the keys to an AI. Should you be worried?

    In mid-July 2026, 1Password launched an integration allowing Claude to connect to your accounts without your passwords ever reaching the model. We explain the mechanism, why it is genuinely well designed, and why it does not address the most important question: what the agent does once connected.

Other topics in “Themes”

All the other topics

Every article, month by month →