The Comunicano for Friday, October 25th 2024

Here Comes The Weekend

Ah, progress—the great tidal wave sweeping us into the future! And what better emblem of progress than AI, that modern marvel poised to revolutionize everything from trading stocks to settling old scores over social media?

Take cybersecurity, for instance. Companies are charging ahead with AI projects, but, funny thing, nearly half of them are leaving their security teams at the gate. Cyber threats? Oh, they’ll just wave from the sidelines. After all, who needs oversight when algorithms can handle… almost everything, right? Like the EU’s AI Act trying to wrangle order out of tech chaos—one day, half the teams might actually read it.

Then, there’s Reddit, which finally raised its voice against the tech giants. They’ve been feeding off Reddit’s user-generated data like it’s a free buffet, and Reddit’s CEO, Steve Huffman, has decided it’s high time to start charging for the entrée. Kudos, Reddit! All those memes and threads are worth a little more than Google’s back pocket change. That’s happening while Perplexity’s battling old-school media over “public” information. And Meta? They’ve struck a handshake deal with Reuters—news straight from the source, finally with permission!

And just when Wall Street looked like it might have peaked on sophistication, here comes Morgan Stanley’s "AskResearchGPT"—a chatbot making bankers sound even more... authoritative. This AI, we’re told, will churn through troves of research, which, let’s be real, is just code for, "Now bankers can spend even *less* time doing their own research." Bravo!

On a grander scale, OpenAI’s "Orion" may or may not be gracing us this December. Sam Altman says it’s all 'fake news'—and these days, who’s to argue? But should it appear, this AI promises to be 100 times the power of GPT-4. Let’s hope it comes with an equally mighty instruction manual, and maybe even a conscience.

And just for fun, Apple decided to keep us guessing, as they often do. The latest iOS beta is adding some streamlined AI interactions. No muss, no fuss, just ChatGPT in your pocket, waiting for permission—because even in the digital age, it seems some things still require a ‘may I?’

So, cheers to AI—the gift that keeps giving, asking, and maybe one day… explaining.

Andy Abramson

AI Watch

No Security—A recent ISACA survey highlights that nearly half of businesses exclude cybersecurity teams from AI development and deployment, despite AI’s growing role in threat detection and response. This gap undermines security efficacy as cyber threats escalate, leaving crucial governance and oversight missing from AI integrations. While regulations like the EU’s AI Act push for more governance, only 35% of security professionals are involved in AI policy discussions, amplifying risks. Including cybersecurity early in AI planning is essential to ensure comprehensive protection and regulatory compliance.

Where The Humans Are—Reddit CEO Steve Huffman, speaking at the *Wall Street Journal*’s Tech Live event, highlighted Reddit's position in the AI ecosystem, emphasizing its value as a source of human-generated data crucial for training AI. While Reddit licenses data to firms like Google, paying $60 million annually, Huffman criticized tech giants that use Reddit's data without consent, advocating for fair compensation. He underscored Reddit's openness to data partnerships, aiming to balance accessibility with protection of its resources. Reddit’s recent policies reflect a commitment to ethical data sharing while supporting advancements in AI through structured agreements.

Talk To A Broker—Morgan Stanley has introduced an OpenAI-powered chatbot, expanding the use of generative AI tools to its investment banking and trading divisions. This new assistant, known as "AskResearchGPT," enables staff to retrieve insights from extensive company research quickly. By leveraging OpenAI's language model, Morgan Stanley aims to improve productivity and streamline information access across teams, a move that reflects broader AI adoption trends in finance. This tool’s rollout follows extensive testing within their wealth management sector and represents Morgan Stanley’s push to integrate advanced AI into core business functions.

Two Sides To Every Story—Perplexity responds to lawsuits by the Wall Street Journa* and New York Post, characterizing these as part of a wider media industry resistance to generative AI tools. They argue that media companies aim to impose control over publicly available facts, which runs counter to Perplexity's mission of open access to information. Emphasizing transparency, Perplexity notes it sources and cites material, distinguishing itself from competitors. They see the lawsuit as adversarial yet shortsighted and highlight their revenue-sharing initiatives with other publishers as a model for cooperation. Perplexity pledges to defend its role in AI-driven knowledge sharing. Read more in Variety.

Meta Watch

Let’s Make a Deal—Meta has partnered with Reuters in a new multi-year agreement to bring real-time news updates to its AI chat functions across platforms like Facebook, Instagram, WhatsApp, and Messenger. Starting with U.S. users on October 25, the Meta AI assistant will utilize Reuters’ news to answer questions on current events, linking directly to Reuters content for additional details. While this marks Meta’s re-engagement with news content following its reduced emphasis on political and news features, it's part of a controlled approach to integrate reliable sources and meet user demand for credible information.OpenAI Watch

No Safety Patrol—Miles Brundage, OpenAI’s former AGI Readiness advisor, has departed from the company, voicing concerns about both OpenAI’s and the industry’s lack of preparedness for the advancement of artificial general intelligence (AGI). He noted that while there is internal acknowledgment of these gaps, the fast pivot towards commercialization has sidelined safety priorities. Brundage’s resignation follows other notable exits from OpenAI, hinting at a growing internal debate about balancing rapid innovation with necessary safeguards. He plans to continue advocating for AI safety in the nonprofit sector, stressing that societal readiness for AGI requires urgent, cross-sector collaboration.

Project Orion—OpenAI plans to release its highly anticipated AI model, codenamed Orion, in December 2024. Orion, based on OpenAI's new GPT-5 "Strawberry" architecture, is expected to significantly advance AI reasoning, planning, and precision, addressing key limitations of previous models. Initially, OpenAI will provide early access to select enterprise partners, such as those on Microsoft's Azure platform, before a broader public release, potentially on December 25, 2024. The model's release underscores OpenAI's strategy to meet increasing demands for AI in complex, high-stakes applications across industries like finance and healthcare.

Not So Fast—OpenAI's CEO Sam Altman recently addressed rumors about a potential December release of the "Orion" model, expected by some to be branded as GPT-5. Following speculation from several sources, Altman dismissed these reports as "fake news" on social media, responding to the increasing buzz around a high-powered model that could purportedly offer up to 100 times the capabilities of GPT-4. Though OpenAI has not confirmed details, Microsoft, a major OpenAI partner, has been preparing to host Orion on Azure for select enterprise partners, suggesting a more limited rollout if it does materialize this year. These comments come amidst heightened interest in OpenAI's next steps, especially given the company's ongoing developments with its "Strawberry" technology, which focuses on improved AI reasoning and accuracy. Industry observers see the Orion model as a significant evolution in OpenAI's trajectory toward more robust AI applications, though official details remain sparse.

Apple Watch

No Time To Chat—iOS 18.2’s beta adds options to disable ChatGPT permission prompts in Apple Intelligence, streamlining interactions with Siri. Users can bypass permissions except when sharing files. Commands starting with “Ask ChatGPT” now route directly to OpenAI's model, which can generate images and original text. The integration doesn’t require an OpenAI account, though linking one is optional. This update is currently limited to the iPhone 16/16 Pro and iPhone 15 Pro/Pro Max models.

Notion Watch

You’ve Got (More) Mail—Notion has introduced a new email client at its “Make with Notion” event, signaling a significant expansion of its platform. This feature aims to integrate email more seamlessly into users' existing workflows, allowing them to manage tasks, projects, and communications in a single, unified interface. The email client is expected to complement Notion's other tools, including databases, calendars, and project management features, thus reducing the need for separate applications for communication and organization. The move reflects Notion's ongoing development toward a multi-product ecosystem to boost productivity and collaboration.

LinkedIn Watch

You’re Verified—LinkedIn has verified 55 million users for free since April 2023, and verified badges will start appearing in user feeds by November. This initiative aims to enhance trust on the platform, especially against fake accounts, which LinkedIn says it intercepts 99% of the time. The verification process includes options like company email authentication or using government IDs through third-party services. Verified profiles reportedly experience higher visibility and engagement, aligning with LinkedIn’s goal of building a trustworthy digital community.

Social Watch

A Messy Divorce—The divorce of influencers Kat and Mike Stickler highlighted the complexities of dividing digital assets like their joint TikTok channel, *MikeAndKat*, which had millions of followers. While Kat took control of the TikTok account, growing her following to over 10 million, Mike kept the now-defunct YouTube channel. For the 27 million U.S. influencers, the valuation of social media accounts in divorces can be challenging, especially when future earning potential is at stake. Couples face unique financial risks, as audiences may dwindle post-split, underscoring the need for careful asset planning and prenuptial agreements in the influencer economy.