Mythos found hundreds of important SharePoint vulnerabilities in weeks, demonstrating how AI-driven discovery could overwhelm vendor capacity and destabilize long-standing approaches to remediation
This post is a tribute to Om Malik’s distinctive style and his impact on my photography.
The UK's NCSC has endorsed passkeys as the default authentication standard, marking a significant shift in cybersecurity practices.
The rapid diffusion of deepfake tools highlights a familiar pattern: technology scales behaviours that society has yet to meaningfully confront or address.
My time in Paris over the 2026 New Year is held here as snippets of light and movement, and reflect some of my efforts to return to myself after an extended period of self-alienation.
My time in Montevideo last year comes together here as snippets of light, people, and fleeting moments rendered in monochrome.
While photography contests aren't universally respected, they can serve as a forcing function by challenging us to critically review our work and what it may be communicating.
What happens when a photographic style—grounded in philosophy—is reduced to a selectable camera profile?
Novelli et al’s paper productively explains how regulatory sandboxes must do more than support innovation: they structure oversight, maintain accountability, and ensure that experimentation strengthens, rather than weakens, regulatory regimes.
Leaders need to continually learn and adapt to enhance their organizations, as highlighted in a recent Harvard Business Review podcast conversation.
A podcast discussion examining the contemporary ransomware ecosystem, including business-like criminal operations, overlapping threat activity, RMM abuse, living-off-the-land attacks, and defensive priorities for smaller organizations.
As ransomware ecosystems fragment and decentralize, declining trust among criminal operators may weaken credibility signals, complicate data-deletion assurances, and affect the ability of ransomware groups to extract payments.
LLM-assisted cybersecurity research is here and making real contributions. But, as Aisle notes, there are questions of how this affects the ability of defenders to ingest and act on vulnerability reports, as well as broader questions of whether adversaries will also use here capabilities to become more capable.
Khan and Jancik propose a structured framework for assessing Canadian AI sovereignty, grounded in policy rationale, competitiveness, and scale. Their focus on technical capacity, however, may elide the ways in which regulators are (now) beginning to shape AI technologies’ adoption.
Nilay Patel recently interviewed Bridget McCormack, former chief justice of the Michigan Supreme Court and now head of the American Arbitration Association (AAA), about the AAA’s new AI-assisted arbitration platform and the broader role AI might play in legal dispute resolution. It was a rich conversation, touching on the design and mechanics of the AI [ ]
This essay examines how the shift from an attention economy to a velocity economy—accelerated by LLMs and agentic systems—may compress decision cycles, expand systemic risk, and impose pressures on institutions as they seek to operate in a velocity economy while retaining legitimacy and social trust.
As platforms move into a “trust-graph” era shaped by generative AI, the redefinition of authenticity, identity, and authority is emerging as a central technology policy challenge for 2026.
Managed to walk into — and out of — a Leica store and only bought a pin. I really wanted to get the used Q2 Monochrome that was ‘only’ 4750 euros. Though I did buy the Leica Academy inspiration book, online and used, after leaving. But I refuse to count it because it’s a published-based [ ]
Over the past week there’s been heightened concern about how LLMs can be used to facilitate cyber operations. Much of that concern is tightly linked to recent reports from Anthropic, which are facing growing criticism from the security community. Anthropic claimed that a threat actor launched an AI-assisted operation which was up to 90% autonomous. [ ]
Researchers found that minimal data poisoning can implant backdoors in large language models, requiring only 250 malicious documents for a 13 billion parameter model. This vulnerability applies to various model sizes, raising concerns about the security of AI systems. Enhanced precautions are needed for safe AI deployment in high-risk environments.