The week of Sept. 7 - 11: What happened, what matters, what's next

Several AI researchers from Anthropic and Google left those companies, sounding an alarm that AI could be on a trajectory that leads to disaster for humanity. This comes after another high-profile warning from Microsoft co-founder Bill Gates who recently published an essay about the potential economic, social and political dangers of AI. The rate at which AI advances and the speed with which AI can target vulnerabilities is a top concern for CIOs and CISOs. Meta also released a new AI agent this week, which demonstrated security vulnerabilities during internal testing.

Read on to catch up on some of the biggest AI developments this week — and what they mean for CIOs.

News Highlights Sept. 7 - 11

AI researchers exit Anthropic and Google

What happened: Two AI researchers from Anthropic and Google DeepMind left their respective companies after Anthropic AI researcher Jacob Coxon quit this week. Joe Benton and Josh Engels, AI safety researchers from Anthropic and Google, respectively, left over concerns shared by Coxon that advanced AI could become uncontrollable and detrimental to humans.

Related:CIOs winning the AI race know when to slow down

Why it matters: It's essential for CIOs and CISOs to establish guardrails and policies for AI ahead of implementation at their organizations. Data privacy and security can't be an afterthought, as demonstrated by the vulnerabilities exposed by Anthropic Mythos and OpenAI's security breach of Hugging Face. Establishing guardrails isn't a one-and-done exercise; as models evolve, guardrails and policies must evolve with them.

Meta unveils new AI agent

What happened: Meta Platforms released a free personal AI agent dubbed Muse; power users can opt for subscription plans of $20 or $100 a month. While competitors such as Anthropic and OpenAI have focused on coding and enterprise applications, Meta's agent is built for consumers.

Why it matters: Despite its target audience, Muse raises questions about shadow AI and privacy for enterprises. Meta's Chief AI Officer Alexandr Wang told the Wall Street Journal that the company is concerned about the potential for AI agents operating outside of express parameters, but that it has established guardrails to ensure agents operate independently of each other to restrict potential sharing of private data. However, Meta employees identified security issues when testing Muse, with some reporting that the agent uploaded "sensitive information without permission," according to Reuters. Still, if adoption of consumer AI agents takes off, CIOs and CISOs will need to ensure they have the right AI guardrails and security policies in place to prevent outside agents from infiltrating the enterprise.

Related:Gartner: CIOs must balance AI value with risk — and fast

Apple's pricey entry into the foldable smartphone market

What happened: Apple is finally releasing the company's first foldable smartphone, the iPhone Duo. Competitors Samsung and Huawei previously released tri-fold smartphones, but the iPhone Duo uses only a single hinge to unfold from a passport size into a single 7.6-inch screen. The device has a big price point starting at $1,999 and will be generally available on Oct. 23.

Why it matters: With access to a larger screen, enterprise employees could swap out their iPads and corporate smartphones for a single device. The larger display's split view option could help employees access multiple applications at once, and the device is compatible with Apple Pencil to make notetaking simpler. Still, the high price point and smaller, less durable screen than the iPad could keep CIOs from going all-in on the iPhone Duo.

NASA and IBM release open source mapping technology

What happened: NASA and IBM have been working together to develop an open source AI model for lunar exploration, reported The Register. The duo curated different types of data formats to develop the AI model that can be used to analyze the Moon's terrain, identify lunar ice deposits and examine lunar volcanoes and craters.

Related:SPIFFE can identify your AI agents. But can it control them?

Why it matters: Open source models, which can be deployed on private clouds or on-premises infrastructure, provide CIOs with a cost-effective avenue for experimentation and proper fit while avoiding vendor lock-in. CIOs and their teams also have more freedom over fine-tuning by using their own training configurations. Open models, however, carry their own risks such as racking up costs if the project is left unchecked and being more prone to security threats compared with proprietary counterparts.

From InformationWeek: Worth the read

You'll find a wealth of insight and analysis on the InformationWeek site, but here are a few curated picks from the week that should be at the top of your reading list.

When is a business really recovered from a cyberattack?
Boston Scientific is still dealing with the fallout from a cyberattack that hit its systems and disrupted manufacturing, order processing and shipping more than two weeks ago. Still, getting systems back online isn't business recovery. Senior Editor Madeleine Streets examines how CIOs measure whether both critical operations and revenue have actually been restored.

John Deere harvests data insights with new AI technology
The company is helping farmers uncover new efficiencies in farming. Its latest addition of a generative AI assistant is already working on harvest schedules. Senior Editor Kelsey Ziser caught up with CTO Jahmy Hindman to discuss why John Deere launched JD and how farmers can use it to better understand their data and operations.

InformationWeek Podcast: Handling vendor-triggered compliance issues
In this episode of the InformationWeek Podcast, Anthony Guethert, CTO of Lumanity; Omer Cohen, CSO at Descope; and Christopher Squier, CISO North America at Sodexo, join Senior Editor Joao-Pierre Ruth to share perspectives on ways compliance issues might be mitigated.

Why observability has become essential to the CIO's job
InformationWeek's Carrie Pallardy spoke with Fernando Castanheira, CIO at Riverbed Technology, and Eric Johnson, CIO at PagerDuty, about how they use observability, how it has changed over the course of their careers and how AI changes the way CIOs anticipate and react to problems.

Coming up: Dates to watch

Sept. 24-25 — The Momentum AI Austin event will focus on how executives can "scale AI, manage risk, drive organizational change, and unlock measurable outcomes," according to the event's website. Speakers include executives from Walmart, ADP, AT&T, BNY, FedEx, Warner Bros, Unilever and more.

Sept. 23-30 — Among the large tech and enterprises announcing earnings results this month are General Mills (Sept. 23), Costco Wholesale Corporation (Sept. 24), BlackBerry Limited (Sept. 24), CarMax (Sept. 29), IDT Corp. (Sept. 29) and Micron Technology (expected Sept. 30).

Oct. 5-8 — Cisco's WebexOne event will be held in Austin, Texas and virtually this year. The agenda includes keynotes and sessions on the "intelligent workplace," AI innovations and personalizing the customer experience with AI.

How is your organization implementing guardrails and security policies for AI deployments? Let us know at [email protected].