This week’s information introduced some telling examples of how rapidly AI is evolving from a know-how experiment right into a sensible device, with materials penalties for a way corporations construct, govern and safe their programs.
In simply the previous few days, the White Home moved to determine voluntary safety testing for superior AI fashions with notable exceptions; Google reshuffled its AI management as a number of longtime researchers departed; and JPMorgan CEO Jamie Dimon launched a cross-industry effort to sort out AI dangers.
On the identical time, a hacking marketing campaign concentrating on dozens of main monetary corporations confirmed that a few of the most persistent enterprise safety issues stay decidedly acquainted. On this case, attackers impersonated IT assist desks and tricked staff into handing over credentials and multifactor authentication (MFA) codes. It is a reminder that IT leaders cannot let novelty distract from the basics.
Listed here are the important thing developments price watching this week:
Information Highlights Aug. 3-7
White Home excludes open-weight AI fashions from voluntary safety testing
What occurred: President Donald Trump’s administration instructed main AI corporations this week that its new voluntary federal cybersecurity testing program won’t cowl open-weight fashions, together with Meta’s Llama and Nvidia’s Nemotron. Representatives from Meta, Anthropic, Google, Nvidia and OpenAI all met with White Home officers to debate the framework.
Why it issues: The choice provides a concrete level of stress to the controversy over how AI programs needs to be examined and overseen. For CIOs, it additionally highlights a tradeoff they more and more face when selecting AI fashions: open programs can supply better management and transparency however will not be topic to the identical testing and safeguards as proprietary fashions. How the federal method develops may affect these selections.
Google reshuffles AI management as longtime researchers go away
What occurred: Google introduced that DeepMind CEO Demis Hassabis will transfer into a md and chief scientist position, with CTO Koray Kavukcuoglu taking on day-to-day management. A number of outstanding AI researchers, together with Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le, additionally introduced plans to depart Google and kind a brand new analysis enterprise.
Why it issues: The adjustments supply a glimpse into the organizational stress on corporations competing to show AI analysis into merchandise. Google has invested closely in consolidating its AI efforts beneath DeepMind, whereas making an attempt to maintain tempo with rivals and commercialize its know-how. The departures increase questions on Google’s inside politics, in addition to how the corporate will stability long-term analysis with the calls for of a quickly altering AI market.
JPMorgan’s Dimon launches cross-industry effort to handle AI dangers
What occurred: JPMorgan Chase CEO Jamie Dimon mentioned this week that the financial institution is increasing its Alliance for Vital Infrastructure to handle rising AI dangers, with greater than 40 corporations throughout monetary providers, power, telecommunications, transportation and different industries approached about becoming a member of. The initiative will concentrate on sharing details about AI use and dangers and creating safeguards, working alongside the Trump administration.
Why it issues: AI threat is changing into a priority past know-how corporations and authorities regulators. The involvement of corporations throughout essential infrastructure additionally factors to a broader problem for CIOs: AI programs are more and more being deployed throughout enterprise and operational environments, making their dangers tough to isolate inside an IT or safety division.
Hackers goal main monetary corporations by impersonating IT assist desks
What occurred: A hacking group focused dozens of main U.S. monetary corporations, together with private-equity corporations, funding managers and monetary exchanges, by impersonating IT help-desk staff. Attackers used spoofed cellphone numbers and pretend web sites to acquire passwords and multifactor authentication codes, whereas Google recognized greater than 200 personalised phishing domains linked to potential targets over 5 weeks.
Why it issues: The marketing campaign reveals how attackers are concentrating on the processes round identification programs slightly than making an attempt to defeat the know-how itself. Assist desks can present a path round MFA as a result of official account-recovery procedures usually depend upon human judgment. For CIOs and CISOs, that places worker verification and identity-recovery processes alongside technical controls as potential factors of failure.
From InformationWeek: Definitely worth the learn
You may discover a wealth of perception and evaluation on the InformationWeek web site, however listed below are a couple of curated picks from the week that needs to be on the prime of your studying record.
At Black Hat 2026, safety leaders go deeper to get forward
This week’s Huge Tech Story options insights from a number of Black Hat attendees in regards to the position of AI on this yr’s occasion and the way the cybersecurity {industry} is shifting previous the hype to concentrate on significant deployments.
InformationWeek Podcast: Coping when inside metrics go off target
This newest episode sees host and senior editor Joao-Pierre Ruth joined by executives from Sage and Daylit to debate how they keep away from letting questionable metrics derail their optimization plans.
CIOs can measure AI spend. Proving its worth is the exhausting half
Contributing author Stephanie Overby digs into the query of ROI on AI initiatives, exploring how IT leaders can measure and display outcomes to government boards.
Your AI brokers will not fail. Your processes will
On this Pam Baker story, the contributing author seems to be on the underlying points that may derail agentic AI deployment and explains why up to date course of design is crucial for achievement.
Arising: Dates to look at
Aug. 9–11 — XChange August gathers IT channel leaders and resolution suppliers in Nationwide Harbor, Md., specializing in managed providers evolution, enterprise AI integration and shifting vendor ecosystem partnerships.
Aug. 12 — U.S. July Inflation (CPI) report is launched by the Bureau of Labor Statistics, offering key macroeconomic information on pricing pressures that immediately affect tech capital expenditure planning and enterprise price range flexibility.
Aug. 12 — Cisco Methods studies fourth-quarter earnings, serving as a serious barometer for enterprise networking demand, cloud infrastructure spending and core IT {hardware} budgets.
From AI governance debates to new approaches for managing AI threat, this week emphasised that there’s a lot extra to efficient AI deployment than simply choosing the proper mannequin. How is your group balancing AI innovation with safety and oversight? Tell us at [email protected]
