CDP Institute

Meta’s WhatsApp detected a hack targeting 90 users in two dozen countries. Meta said that Israeli spyware company Paragon Solutions is responsible and has sent a cease-and-desist letter. The targets apparently included journalists and “members of civil society.”

OpenAI has a new AI agent called Deep Research to support, well, deep research. OpenAI says it can conduct multi-step research tasks, synthesizing hundreds of online sources to create a comprehensive report at the level of a research analyst. It will support research into complex purchases, possibly including martech software.  There’s no truth to the rumor that it answers every question with, “It depends.”

Meltwater’s 640-page compendium of global digital information highlights a “low stakes social” trend: audiences are asking social media to provide some light relief from the emotional turmoil of real life. On average, users are on social more than two hours each day (out of more than six hours on the internet in general). Marketers’ biggest challenge in handling social media is team bandwidth.

AI vendor Anthropic has developed a new technique to protect against universal jailbreaks, jailbreaks that can break all of a model’s defenses. The technique focuses on Constitutional Classifiers, trained on synthetically generated data, that filter the overwhelming majority of jailbreaks. Anthropic claims minimal over-refusals and no large increase in compute overhead.

Content discovery platform Outbrain has completed an almost $1 billion merger with programmatic ad platform Teads. The combined solutions will operate under the Teads name. This comes several years after the high-profile merger between Outbrain and Taboola fell apart.

The tech industry group Netchoice has sued to block legislation that would require businesses to assess whether new features would harm minors and restrict their ability to collect data from minors. The legislation was enacted in Maryland last May. It is similar to the Californian Appropriate-Design Code, also held up by the courts.

Faced with competition from China’s open source AI model DeepSeek, OpenAI CEO Sam Altman has admitted that his company was on the “wrong side of history” for not going open source itself. Meanwhile, DeepSeek has been found vulnerable to “jailbreaking” and prone to hallucinations, while bad actors in Iran and elsewhere have become heavy users of Google Gemini. Any need for government oversight here?

Hackers new to cybersecurity have found themselves targeted by a trojanized version of XWorm RAT builder. The malware has been spread through GitHub, Telegram and file-sharing programs and has compromised over 18,000 devices. Targets are mainly found in the U.S., Russia, India, Ukraine, and Turkey.

In an experiment from Google’s search labs, AI agents are offered who can “wait on hold for you.” The Ask for Me tool will call local businesses on your behalf to set up appointments or collect information on product availability and pricing. Ask for Me can be enabled on Search Labs now and will maybe put a lot of concierges out of work.

Marketing and advertising holding group Dentsu has announced the launch of Adobe GenStudio Dentsu+, combining Adobe’s AI content creation platform with Dentsu’s services and the Dentsu-owned Merkury ID platform. The result is described by Dentsu as a “content supply chain platform” for its clients. This is further evidence that the larger agencies and holding companies are becoming, in part, tech companies too.

Perhaps because Target had seemed sincere in its support for minority groups including the LGBTQ+ community, the announcement that it would end its DEI initiatives sparked anger and calls for a boycott. Other brands, such as Walmart, are taking similar negative steps. Some 98% of Costco shareholders, however, voted down a proposal to conduct a risk analysis of its DEI program.

The Identity Theft Resource Center reports that 2024 data breaches reached the second highest level since they started tracking them in 2005. The number (3,158) fell just 44 short of 2023’s record level. Because the total includes six “mega-breaches” involving 100 million to 560 million victims, the total number of notices served to individuals quadrupled from 400 million to 1.7 billion.