Fannie Mae & Freddie Mac Tighten Lending Rules

Fannie Mae and Freddie Mac will enforce stricter lending rules for commercial property.

New regulations will require lenders to verify borrowers' financial information. They must confirm borrowers have enough cash and verify the source of their funds.

Currently, lenders take a lax approach, often trusting submitted financial statements. However, the rise in high-interest rates has led to a fall in property values and increased scrutiny of loan applications.

Fannie Mae and Freddie Mac own or guarantee about 40% of the $2.2 trillion multifamily mortgage debt in the US.

Source: Read More

Brokers Outage Amid Market Plunge

Yesterday, Charles Schwab, Fidelity, Robinhood, and Vanguard, faced outages. These outages coincided with a steep decline in stock indexes, causing frustration among investors.

Reports indicated over 15,000 users experienced issues with Schwab’s website, while Fidelity and Vanguard logged more than 3,600 and 2,800 complaints, respectively.

Users struggled to log in and execute trades during the panic sell-off. Schwab and Fidelity managed to resolve their outages by midday.

The disruptions heightened concerns about the platforms' ability to handle high trading volumes during extreme market swings.

Source: Read More

Google Loses Antitrust Suit

A federal judge ruled that Google violated antitrust laws by maintaining a monopoly on search engines.

Judge Amit Mehta stated that Google paid $26 billion to secure its position as the default search engine on smartphones. This strategy effectively blocked competitors from entering the market.

With about 90% of online searches and 95% of smartphone searches under its control, Google’s actions limited consumer choice.

The judge labeled Google a monopolist. Future trials may explore solutions, including banning Google's payments to smartphone manufacturers.

Source: Read More

AI "Inbreeding": Whats is Model Collapse?

"Model collapse" arises when AI systems are trained using content created by other AI. Researchers believe this can lead to declining performance over time.

As AI-generated content floods the internet, the quality of AI tools will suffer.

Products like ChatGPT rely heavily on data scraped from the web, often including AI-generated content. This creates a cycle of "inbreeding," where models reproduce the same information, losing uniqueness and depth.

Moreover, examples illustrate how recursive training can lead to nonsensical outputs. A model initially focused on European architecture ended up producing random statements about jackrabbits.

Source: Read More