About I2C communication of AD5933 Both SDA and SCL are connected with 10KΩ pull-up resistors, and bypass capacitors are not attached. During the evaluation test, when I briefly shorted both ends of the pull-up resistor

It is not a good idea to short the pull-up resistor to VDD. The SDA pin is open drain and current will exceed the maximum rating if the AD5933 tries to output a low level.

If this is a possible failure condition, the SDA pin should have a series resistor to ensure that maximum ratings are not exceeded.

It's difficult to tell if the misbehaviour of the AD5933 is due to the current overload or to the bit pattern seen during the error condition.

It is not a good idea to short the pull-up resistor to VDD. The SDA pin is open drain and current will exceed the maximum rating if the AD5933 tries to output a low level.

If this is a possible failure condition, the SDA pin should have a series resistor to ensure that maximum ratings are not exceeded.

It's difficult to tell if the misbehaviour of the AD5933 is due to the current overload or to the bit pattern seen during the error condition.

The same phenomenon occurred even without shorting with tweezers.
I would like to know the possible causes.

Also, could you please tell me what it means that the malfunction of the AD5933 is caused by a current overload?

thank you.

That points to some corruption or hiccup in the I2C bus that sets the slave out of sync with the master. You can try to get the slave out of this condition by issuing 10 SCL cycles and checking that both lines return to high state.

If the overload condition takes long enough, the transistor at the SDA output can get damaged and probably keep the line permanently low, preventing any communication on the bus. Another possibility is a latch-up caused by current injected in the SDA pin when shorting the resistor. Latch-up causes a semiconductor junction to enter conduction mode with high current and it can only be cleared with a power cycle.